1: \documentclass[12pt,preprint]{aastex}

2: 

3: %% The text in this version corresponds to the proofed version of the

4: %% paper.

5: 

6: 

7: \newcommand{\ud}{\mathrm{d}} \newcommand{\feh}{\mathrm{[Fe/H]}}

8: \newcommand{\fe}{\mathrm{[Fe/H]}}

9: %\newcommand{\bv}{$(B-V$}

10: \newcommand{\avevR}{\langle v_R \rangle}

11: \newcommand{\avevphi}{\langle v_\phi \rangle}

12: \newcommand{\avevz}{\langle v_z \rangle}

13: \newcommand{\kms} {\mathrm{km \ s^{-1}}}

14: \newcommand{\lz}{$L_z$}

15: \newcommand{\lperp}{$L_{\perp}$}

16: 

17: \shorttitle{Local Halo Star Streams}

18: \shortauthors{Kepley et al.}

19: 

20: \begin{document}

21: 

22:   \title{Halo Star Streams in the Solar Neighborhood}

23:   \author{Amanda A. Kepley}

24:   \affil{Department of Astronomy, Case Western Reserve University,

25:   Cleveland, OH 44106, U.S.A.; and Department of Astronomy, University

26:   of Wisconsin--Madison, 475 North Charter Street, Madison, WI 53706, U.S.A.

27: \email{kepley@astro.wisc.edu}}

28: \author{Heather L. Morrison}

29:   \affil{Department of Astronomy, Case Western Reserve University,

30:   Cleveland, OH 44106, U.S.A. \email{ heather@vegemite.case.edu}}

31:   \author{Amina Helmi}

32:   \affil{Kapteyn Astronomical Institute, University of Groningen, P.O. Box 800, 9700 AV Groningen, Netherlands \email{ahelmi@astro.rug.nl}}

33: \author{T.D. Kinman}

34: \affil{National Optical Astronomy Observatory, P.O. Box 26732, Tucson

35:   AZ 85726, U.S.A.  \email{tkinman@noao.edu} }

36:   \author{Jeffrey Van Duyne}

37:   \affil{Department of Astronomy, Yale University, P.O. Box 208121,

38:   New Haven, CT 06520, U.S.A. \email{vanduyne@astro.yale.edu} }

39:   \author{John C. Martin} 

40:   \affil{University of Illinois at Springfield, 1 University Plaza, MS

41:   HSB 314, Springfield, IL 62703, U.S.A. \email{jmart5@uis.edu}}

42:   \author{Paul Harding}

43:   \affil{Department of Astronomy, Case Western Reserve University,

44:   Cleveland, OH 44106, U.S.A. \email{harding@dropbear.case.edu}}

45: \author{John E. Norris and Kenneth C. Freeman}

46: \affil{Research School of Astronomy and Astrophysics, Australian

47:   National University, Private Bag,

48: Weston Creek PO, 2611 Canberra, ACT, Australia \email{jen@mso.anu.edu.au,kcf@mso.anu.edu.au}}

49: 

50: \begin{abstract}

51: 

52: We have assembled a sample of halo stars in the solar neighborhood to

53: look for halo substructure in velocity and angular momentum space. Our

54: sample (231 stars) includes red giants, RR Lyrae variable stars, and

55: red horizontal branch stars within 2.5 kpc of the Sun with [Fe/H] less

56: than --1.0. It was chosen to include stars with accurate distances,

57: space velocities, and metallicities as well as well-quantified

58: errors. With our data set, we confirm the existence of the streams

59: found by Helmi and coworkers, which we refer to as the H99

60: streams. These streams have a double-peaked velocity distribution in

61: the $z$ direction (out of the Galactic plane). We use the results of

62: modeling of the H99 streams by Helmi and collaborators to test how one

63: might use $v_z$ velocity information and radial velocity information

64: to detect kinematic substructure in the halo. We find that detecting

65: the H99 streams with radial velocities alone would require a large

66: sample (e.g., approximately 150 stars within 2 kpc of the Sun and

67: within 20$^\circ$ of the Galactic poles). In addition, we use the

68: velocity distribution of the H99 streams to estimate their age. From

69: our model of the progenitor of the H99 streams, we determine that it

70: was accreted between 6 and 9~Gyr ago. The H99 streams have

71: [$\alpha$/Fe] abundances similar to other halo stars in the solar

72: neighborhood, suggesting that the gas that formed these stars were

73: enriched mostly by Type II supernovae. We have also discovered in

74: angular momentum space two other possible substructures, which we

75: refer to as the retrograde and prograde outliers. The retrograde

76: outliers are likely to be halo substructure, but the prograde outliers

77: are most likely part of the smooth halo. The retrograde outliers have

78: significant structure in the $v_\phi$ direction and show a range of

79: [$\alpha$/Fe], with two having low [$\alpha$/Fe] for their [Fe/H].

80: The fraction of substructure stars in our sample is between 5\% and

81: 7\%. The methods presented in this paper can be used to exploit the

82: kinematic information present in future large databases like RAVE,

83: SDSSII/SEGUE, and {\em Gaia}.

84: 

85: \end{abstract}

86: 

87: \keywords{methods: statistical, Galaxy: halo, Galaxy: kinematics and

88: dynamics, solar neighborhood }

89: 

90: \section{Introduction}

91: 

92: The kinematics of stars in the Galaxy provide information about its

93: structure and formation history. Early Galaxy formation

94: models, most notably those by Eggen et al. (1962),

95: postulated a monolithic collapse of gas and dust, with a large cloud

96: of gas collapsing and forming stars. Later, \citet{sz78} suggested

97: that the Galaxy was formed through a series of accretion events, with

98: smaller proto-galaxies coming together to form a larger

99: structure. Their scenario is more in keeping with our current

100: understanding of how structures formed in the early universe in a

101: ``bottom-up'' fashion \citep{sm94,ws00}.

102: 

103: Ibata et al. (1994) discovered that the Galaxy is currently accreting

104: the Sagittarius dwarf spheroidal galaxy. Stars from this system are

105: not yet well-mixed with the rest of the Galactic halo either spatially

106: or in velocity space. Could other halo accretion events be detected if

107: the stars from the accreted object have become well-mixed with the

108: stars in the halo spatially, but not in velocity or angular momentum

109: space?  \citet[][hereafter H99]{h99}, found evidence in angular

110: momentum space for streams in the solar neighborhood using a sample of

111: 97 metal-deficient red giants and RR Lyrae stars within 1 kpc of the

112: Sun and, using the sample properties, generated a model of the

113: streams.

114: 

115: The purpose of this work is to search for further evidence for

116: substructure in the local Galactic halo and to develop novel ways to

117: detect halo substructure. In order to detect the subtle signs of

118: kinematical substructure, accurate data are needed. We use two

119: different halo star samples: the original H99 sample and a combined

120: data set that includes local giants, red horizontal branch (RHB)

121: stars, and RR Lyrae stars, all selected without kinematic bias. Both

122: samples have full space velocities, distances, and

123: metallicities. Further improvements to the solar neighborhood halo

124: sample will be presented in the next paper in this series,

125: H.L. Morrison et al. (2007, in preparation, hereafter M07), which will

126: focus on the overall properties of the sample and their constraints on

127: the formation history of the local halo, rather than specifically on

128: substructure.

129: 

130: Full space velocities are difficult to obtain for many stars,

131: preventing a complete analysis of their motions like that of

132: H99. However, the H99 models and original data show structure even in

133: their radial velocity distributions, especially when stars in a

134: particular direction, e.g. near the Galactic poles, are considered. In

135: addition, other possible substructures that we discuss in this paper

136: occupy the tails of the angular momentum distribution, and so may be

137: detectable in radial velocities dominated by rotational velocity.

138: Standard statistical tests such as the Shapiro-Wilk test for

139: normality, which is quite sensitive to the behaviour of the tails of

140: the distribution, can be used to detect such deviations. We use the

141: H99 model of the streams as well as a model of a smooth halo to show

142: how streams can be detected in one component of velocity ($v_z$), and

143: we determine the detection limits for this method. We also extend this

144: method to radial velocities for stars in the direction of the Galactic

145: poles.

146: 

147: In recent years, increasingly large and precise data sets of

148: abundances for multiple elements have made it possible to study a new

149: dimension of substructure in the halo: the chemical

150: patterns. Abundances trace the star formation history of the objects

151: that became today's halo.  A good general discussion of the

152: possibilities of this technique is given in \citet{kcf_josh}.  In

153: particular, the ratios of the $\alpha$-elements such as Mg, Ca, and Ti

154: to Fe allow the possibility of distinguishing between a chemical

155: history where enrichment only comes from short-lived massive stars

156: (Type II supernovae), and one where the contribution of Type Ia

157: supernovae need to be considered as well. Type II supernovae

158: preferentially enrich the interstellar medium (ISM) with $\alpha$

159: elements, while Type Ia supernovae produce more iron but need a longer

160: period of chemical evolution. Thus the bulk of the local halo stars

161: have a higher value of [$\alpha$/Fe] than local disk stars, reflecting

162: different formation conditions. However, there are some metal-poor

163: stars that have unusually low values of [$\alpha$/Fe], such as those

164: in dwarf spheroidal galaxies \citep{shetrone01,shetrone03,tolstoy03},

165: a small number of outer halo globular clusters such as Pal 12 and Rup

166: 106 \citep{brown97,brown99}, and some local halo field stars

167: \citep{carney97,king97,nissen97,fulbright2002,stephens2002,inese03,venn04}. For

168: stars from substructures identified via kinematic methods, we

169: investigate whether there are any unusual patterns in [$\alpha$/Fe].

170: 

171: Section \ref{datasets} describes the compilation of the data sets used

172: in this work in detail. Information about our model of the Galactic

173: halo is given in Section~\ref{model} and our results are described and

174: discussed in Section~\ref{results}. Section~\ref{conclusions}

175: summarizes our conclusions.

176: 

177: \section{Data Sets} \label{datasets}

178: 

179: The search for substructure in H99 is particularly powerful because

180: estimates of all three space velocities are available for their

181: sample. This allows the use of angular momentum (approximately

182: conserved in a roughly spherical potential) to isolate stars with

183: similar origins. In the pre-{\em Gaia} era, the need for three space

184: velocities limits us to a relatively small volume surrounding the Sun,

185: where sufficiently accurate proper motions are available.  It is

186: important to note that methods of identifying satellite debris using

187: physical quantities such as angular momentum and energy

188: \citep[e.g.][]{lb2,h99,h06} rely strongly on good distance estimates.

189: (The angular momentum estimate varies as distance squared, since two

190: out of three velocity coordinates are obtained by multiplying the

191: proper motion by the distance.) Thus the accuracy of distance

192: estimates is very important.

193: 

194: \subsection{H99 Data}

195: 

196: The H99 sample stars were selected from the compilations of

197: \citet{bs95} and \citet[][hereafter CY98]{cy98} to have distances less

198: than 1 kpc and [Fe/H] less than --1.6.  This sample contains nearby

199: stars in three evolutionary states: first ascent red giants, RHB

200: stars, and RR Lyrae variable stars. These stars present different

201: challenges for stellar population work. Good velocity estimates are

202: difficult to obtain for RR Lyrae stars, but are relatively

203: straightforward for red giants. This is because RR Lyrae stars show

204: significant velocity variation due to pulsation, while first ascent

205: giants and RHB stars have spectra that lend themselves readily to

206: accurate velocity estimation. Conversely, distance estimates for RHB

207: and RR Lyrae stars are good, but distance estimates for red giant

208: stars less so.  RR Lyrae and RHB stars show little variation in

209: absolute magnitude with metallicity. For example, \citet{mm98}

210: estimated a typical error of 10\% for RR Lyrae distances, and

211: \citet{vivas05} estimate 6\%. First ascent red giants generally have

212: less precise distances, because the position of the giant branch in

213: the color-magnitude diagram depends strongly on metallicity, and so

214: metallicity measurement errors propagate to a larger distance error of

215: order 20\% (see discussions in \citealt{mff90,mo03}). The advantage of

216: smaller distance errors for RHB stars is offset by the difficulty of

217: identifying these stars; in order to separate the subtle differences

218: in gravity between first ascent red giants below the horizontal branch

219: and horizontal branch stars we need photometry from intermediate-band

220: systems like Stromgren or DDO \citep{b80,nbp,att94}. Luckily, many of

221: the stars in the H99 sample were observed by \citet[][hereafter

222: ATT94]{att94} who obtained accurate Stromgren {\it uvby} colors,

223: enabling them to classify stars as either RHB or first-ascent

224: giants. Below we calculate the actual errors on distances, space

225: velocities and angular momenta for the ATT stars, illustrating that

226: these quantities are known remarkably precisely for a sample including

227: red giants.

228: 

229: \subsection{Combined Data Set} \label{combined_sample_sec}

230: 

231: Since H99 was published there have been several improvements on the

232: information available for the solar neighborhood giant and horizontal

233: branch stars. In particular, \citet[][B00 hereafter]{b00} presented a

234: catalog of 2016 stars selected without kinematic bias, and we use

235: preliminary results from a new sample of RR Lyraes with significantly

236: improved radial velocity measurements (M07).

237: 

238: \subsubsection{RR Lyrae Variables}

239: 

240: While the \citet{l94} data set provided accurate [Fe/H] and distance

241: measurements for the local RR Lyrae stars, the velocity measurements

242: were less accurate (typical errors of 30~km~s$^{-1}$).  M07 obtained

243: more accurate velocities for the RR Lyrae stars accessible from the

244: north, observing most stars more than once and using high-quality

245: light curves obtained by T.D.  Kinman (1991, private communication) to

246: correct for the large radial velocity amplitude of each

247: star. Velocities were corrected for pulsation using the synthetic

248: velocity curves of \citet{liu}. Typical velocity errors are

249: 15~km~s$^{-1}$. \citet{mm98} published improved values of proper

250: motion for RR Lyrae stars by averaging {\em Hipparcos} proper motions

251: with accurate ground-based determinations from the USNO ACT reference

252: catalog \citep{ucw98}. We have selected a sample of 96 stars with

253: [Fe/H] less than --1.0 from these data sets. Metallicities for these

254: stars are from \citet{l94,l96}. The distances of stars in the M07

255: sample (derived using the \citet{l94} period-luminosity relation)

256: range from about 0.5 kpc (RR Lyrae itself) to 2.5 kpc. Our sample

257: presently only includes RR Lyrae stars of type ab.

258: 

259: %% \citet{mm98} published improved values of proper motion for RR Lyraes

260: %% by averaging Hipparcos proper motions with accurate ground-based

261: %% determinations from the USNO ACT reference catalog \citep{ucw98}. We

262: %% have selected a sample of stars with [Fe/H] less than --1.0 from these

263: %% data sets. The distances of stars in the M07 sample range from about

264: %% 0.5 kpc (RR Lyrae itself) to 2.5 kpc.

265: %% %% 96 stars with [Fe/H] less than --1.0

266: 

267: \subsubsection{Red giants and RHB Stars} \label{combined_rg}

268: 

269: We have selected a subset of the red giant stars in the B00 catalog

270: for our expanded sample. In order to preserve the distance accuracy so

271: important to the calculation of angular momentum, we critically

272: compared the distance estimates given in the B00 catalog with those

273: given by earlier work such as ATT94. This uncovered some surprisingly

274: large, systematic differences in distance estimates.

275: 

276: Figure~\ref{boo_cy_dist} shows that the B00 and ATT94 distances agree

277: reasonably well for distances greater than 1 kpc. For distances less

278: than 1 kpc, the ATT94 distances are on average a factor of 2 larger

279: than the B00 distances, a remarkably large amount for local,

280: well-studied stars.

281: 

282: Both papers use \bv\ colors and the globular cluster giant branch

283: parameterization of \citet{nbp} to estimate absolute magnitudes for

284: the giant stars. B00 assumed that all giants are on the first-ascent

285: giant branch, while ATT94 used the Stromgren $c1$ index to separate

286: horizontal-branch and asymptotic giant branch (AGB) stars from

287: first-ascent giants.  AGB star contamination is expected to be small

288: because of the short AGB lifetimes, but horizontal-branch stars are

289: more common. Blueward of $(B-V)_0\simeq0.9$, we expect significant

290: numbers of RHB stars to appear in the sample. Assuming that these

291: stars were on the first ascent giant branch, as B00 did, will lead to

292: significant underestimates of their distance. This error will become

293: larger for the bluest stars. Figure \ref{boo_cy_bv} shows that this is

294: indeed the case, with B00 distances being on average a factor of 2

295: smaller than the ATT94 distances for $(B-V)_0=0.6$.

296: 

297: Thus we have restricted our sample to giants from B00 with $(B-V)_0$

298: greater than 0.9, red giants and RHB stars from CY98 with distances

299: determined by ATT94, and RR Lyrae stars from the M07 sample.  We have

300: also restricted the sample to stars with [Fe/H] less than --1.0 and

301: distances less than 2.5 kpc, and eliminated stars with thick disk

302: kinematics (see Section~\ref{streams_in_combined} for details). The

303: $(B-V)_0$ criteria led us to accept 24 red giants and to reject 85 red

304: giants from B00 that had otherwise acceptable distances and

305: metallicities. Fourteen of the accepted stars were later eliminated

306: from the final sample because of their thick disk kinematics (see

307: Section~\ref{streams_in_combined}).  For stars in our sample within

308: 0.5 kpc of the Sun, our distances agree with {\em Hipparcos} distances

309: \citep{hipcat} within the errors. The only exception is HD 135449,

310: whose distance in Table~\ref{combined_sample_table} is twice the

311: distance given by Hipparcos. In future versions of this sample

312: (M07), we plan on using the

313: Hipparcos distances for nearby stars. Metallicities in B00 come from a

314: large variety of sources, but the ATT94 data (which dominate our

315: combined sample of red giants and RHB stars) are significantly more

316: homogeneous, being based either on {\it uvby} photometry (89 stars, an

317: estimated [Fe/H] error 0.16 dex) or accurate spectroscopic abundances

318: from the literature (68 stars).

319: 

320: %% Thus we have restricted our sample to giants from B00 with $(B-V)_0$

321: %% greater than 0.9, red giants and red horizontal branch stars from CY98

322: %% with distances determined by ATT94, and RR Lyraes from the M07

323: %% sample. We have also restricted the sample to stars with [Fe/H] less

324: %% than --1.0 and distance less than 2.5 kpc.

325: 

326: \subsubsection{Error Estimates}

327: 

328: Because the data in our combined sample come from a variety of sources

329: and include stars in different evolutionary stages, it is preferable

330: to calculate the errors on derived quantities individually, since they

331: can vary significantly (see Table 1 and Fig.

332: \ref{meta_fehm10_d25_ang_mom_groups}).  We propagated the known errors

333: on distance, velocity, and proper motion through the calculation of

334: $v_R, v_\phi, v_z$, $J_z$, and $J_{\perp}$ using a Monte Carlo

335: calculation, drawing new values of the input quantities from Gaussian

336: distributions, recalculating the derived values and measuring the

337: standard deviation of their distributions.

338: 

339: For the RR Lyrae and RHB stars, distance errors are relatively small,

340: and have been assumed to be 7\%. For the first ascent red giants in

341: the sample, it is more difficult to quantify errors because the

342: metallicity is the main driver of distance errors and many authors do

343: not quote metallicity errors\footnote{Note that CY98 calculated

344: distance errors simply by assuming that the error on each metallicity

345: value in their sample was 0.16 dex, while ATT94's actual quoted errors

346: vary from 0.01 to 0.31 dex. We have set errors quoted at less than 0.1

347: dex to be 0.1 dex in our calculations.}. We have calculated distance

348: errors individually for stars in the ATT94 sample (where accurate

349: metallicity errors are given, and in fact a significant number of

350: stars have multiple high-dispersion metallicity estimates) and for a

351: small number of red giants, most of which are part of the outlier

352: groups discussed in Section~\ref{streams_in_combined}. To calculate

353: distance errors, we used \bv\ color and the globular cluster giant

354: branch loci of \citet{nbp} with a Monte Carlo calculation of the

355: effect of metallicity and color errors on the estimate of absolute

356: magnitude.

357: 

358: For the five red giant stars not in the ATT94 sample, we made the

359: following assumptions. Four (HD 18710, HD 174578, HD 214925 and CD

360: --68 1881) have [Fe/H] values from \citet{nbp}. \citet{tat94} have

361: shown that the DDO [Fe/H] calibration used for these stars has

362: systematic errors in some metallicity ranges. We have assigned [Fe/H]

363: errors of 0.25, 0.50, 0.20 and 0.50 dex respectively to account for

364: this. BD +30 2282 has metallicity from \citet{hy82}; we have assigned

365: an [Fe/H] error of 0.35 dex for this star. There was also one giant in

366: ATT94 (HD 128279) that was too blue to use the \citet{nbp} distance

367: calibration for the Monte Carlo calculation.  Luckily, it has a {\em

368: Hipparcos} parallax of reasonable accuracy so we have simply used the

369: parallax and its error rather than our other distance error estimation

370: procedure.

371: 

372: \subsubsection{Final Sample}

373: 

374: We have error estimates for distances and derived quantities

375: (velocities and angular momentum) for the great majority of the

376: sample: 210 out of 231 stars.  The combined sample has a median

377: distance of 1.1 kpc, median distance error 7\%, and median metallicity

378: [Fe/H] = --1.7. Median errors on $v_R, v_\phi, v_z$ velocities are 15,

379: 20 and 11 km~s$^{-1}$ respectively, making this sample well-suited for

380: careful investigations of substructure.

381: 

382: Table \ref{combined_sample_table} lists values of distances,

383: metallicities, radial velocities, and galactocentric velocities for

384: the red giants and RHB stars in the combined sample as well as the

385: associated errors for these quantities. The RR Lyrae data will be

386: published in M07.  In Table~\ref{combined_sample_table}, the radial

387: velocities are heliocentric and the $v_R$, $v_\phi$, and $v_z$

388: velocities are relative to the center of the Galaxy on a left-handed

389: coordinate system. This has $R$ pointing away from the center, $\phi$

390: increasing in the direction of Galactic rotation, and $z$ pointing

391: towards the north Galactic pole (NGP). Note that our angular momenta

392: are calculated on this left-handed system in order to directly compare

393: our results with the results of H99. (See the Appendix for more

394: details.)  The velocities were corrected for the motion of the Sun and

395: LSR using $v_{lsr} = 220.0 \ \kms$, $U_{\odot} = -9.0 \ \kms$,

396: $V_{\odot} = 12.0 \ \kms$, and $W_{\odot} = 7.0 \ \kms$ on the

397: standard left-handed coordinate system \citep{blaauw,mb81}. The Sun

398: was assumed to be 8.0 kpc from the Galactic center. See the Appendix

399: for the transformation between the local, Sun-centered coordinate

400: system and the galactocentric coordinate system.

401: %Thick disk stars were

402: %identified and excluded from the sample using their position in a plot

403: %of $J_\perp$ versus $J_z$ (see below for details).

404: 

405: The sample was selected without kinematic bias, but is spatially

406: incomplete: it has fewer stars with low $|b|$ because of the

407: difficulty of identifying these rare halo stars in heavily reddened

408: regions. We show the distribution of distance and $|b|$ in

409: Figure~\ref{lb}. It can be seen that the RR Lyrae sample is more

410: complete in $|b|$ than the giant sample, although neither are fully

411: complete in $|b|$. The giant sample has a lower mean distance than the

412: RR Lyrae sample because of the intrinsic rarity of the RR Lyrae

413: stars. Since all three velocity components can be measured for each

414: star, the spatial incompleteness does not cause a bias in the velocity

415: distribution directly. However, the bias toward high Galactic latitude

416: does select against stars with a flattened space distribution, such as

417: the metal-poor stars in the thick disk \citep{nbp,mff90}. Therefore,

418: we would expect any such stars to be under-represented in our sample.

419: 

420: Since there will be a significant number of binaries in any sample of

421: stars (for example, 17\% in the sample of metal-poor giants of

422: \citealt{carney03}), we need to consider the effect of binarity on our

423: estimates of distance and velocity.  Fortunately, because our sample

424: is composed of giants and horizontal branch stars, the effect of any

425: undetected binaries on the distance estimates will be

426: small. \citet{carney03} studied 91 metal-poor red giants and RHB

427: stars, (68 in common with our sample) and identified 8 spectroscopic

428: binaries. The median velocity amplitude of these binaries is

429: 8~km~s$^{-1}$, which is small compared to our typical error on a U,V

430: or W velocity (10-15~km~s$^{-1}$). While we expect a similar number of

431: undetected spectroscopic binaries in our sample, the effect on our

432: conclusions will be negligible. Similarly, the ``velocity jitter''

433: that is seen in some stars close to the giant branch tip is of an even

434: smaller magnitude (of order 5~km~s$^{-1}$) and will not affect our

435: conclusions either.

436: 

437: 

438: 

439: \section{Model Information} \label{model}	

440: 

441: \citet{hw99} described models used to study the disruption of satellite

442: galaxies by the Milky Way. H99 used these to model the progenitor of

443: the star streams they detected in the solar neighborhood. The remnants

444: of a progenitor with an initial velocity dispersion of 18~km s$^{-1}$

445: and core radius of 0.5-0.65~kpc fit the star streams discovered in

446: H99. Its orbit has an apocenter at 16~kpc, pericenter at 7~kpc, a

447: maximum height above the plane of 13~kpc, and a radial period of

448: 0.4~Gyr. The present paper uses an improved version of the model,

449: which better accounts for the self-gravity of the satellite and has

450: positions and space velocities for $10^6$ particles in the disrupted

451: progenitor from 0~Gyr-13~Gyr.  After 3-5 Gyr, these particles

452: fill much of the space between 7 and 16 kpc from the Galactic center

453: (the amount of debris from the progenitor peaks at the solar circle)

454: and are very well-mixed spatially within $\sim$5 kpc from the Sun.

455: The disrupted progenitor still has some structure in velocity and

456: angular momentum space, however, as can be seen in

457: Figure~\ref{ang_mom_am}.

458: 

459: In addition, we have created a model of a smooth halo. This model

460: allows us to test the null hypothesis that the halo has no

461: substructure. The density distribution of the smooth halo (number of

462: stars per unit volume) is proportional to $r^{-3.5}$ \citep{z85,vivas06}. The

463: velocity distribution of the halo was modeled using a velocity

464: ellipsoid \citep{swar}.  The models were created by randomly selecting

465: points from these distributions. Figure~\ref{ang_mom_smooth} shows one

466: realization of this model in velocity and angular momentum space. The

467: parameters $\avevR$, $\avevphi$, $\avevz$, $\sigma_R$, $\sigma_\phi$,

468: and $\sigma_z$ were determined by finding the average and the standard

469: deviation of the $v_R$, $v_\phi$, and $v_z$ velocity distributions

470: from the combined data set excluding known stream stars.

471: 

472: The standard velocity ellipsoid numbers for the halo

473: \citep[e.g.][]{cb00} were not used because the H99 star streams, in

474: particular, bias the velocity distributions. For example,

475: Figure~\ref{ang_mom_h99} shows that the stream stars have large

476: absolute $z$ velocities. Including these stars in the velocity

477: ellipsoid calculations changes $\sigma_z$ from 84 km s$^{-1}$ to 101

478: km s$^{-1}$.   Table~\ref{smo_params} gives the derived velocity

479: ellipsoid parameters for our combined sample (stars with distances

480: less than 2.5 kpc and $\feh \leq -1.0$, excluding likely thick disk

481: stars), both with and without stream stars.

482: %Non-zero values for $\avevR$, $\avevphi$, and $\avevz$ were used. 

483: In the case of $\avevR$ and $\avevz$, the parameters used are

484: consistent with zero within the errors. However, $\avevphi$ is not

485: consistent with zero. \citet{cb00} find a similar $\avevphi$ for their

486: sample within 1~kpc of the Sun and note that $\avevphi$ decreases when

487: larger distance ranges are included. Note that excluding the star

488: streams from the sample used to determine the velocity ellipsoid also

489: increases the anisotropy of the halo.

490: 

491: %\section{Results} \label{results}

492: 

493: \section{Identifying Structure in the Tails of the Velocity Distribution} \label{results}

494: 

495: \subsection{Method} \label{method}

496: 

497: Using distances and full space velocities of metal-poor red giants and

498: RR Lyrae stars, H99 identified the remnant of a merger between a small

499: satellite galaxy and the Milky Way. Since obtaining proper motions of

500: stars is difficult, another way to identify subtle merger remnants is

501: desirable. Here we explore signatures visible in the tails of the

502: velocity distribution. Figure~\ref{ang_mom_h99} plots the H99 data in

503: velocity and angular momentum space. Its bottom panel plots $J_z$, the

504: component of the angular momentum (per unit mass) out the plane of the

505: Galaxy's disk, versus $J_{\perp} = \sqrt{J_x^2 + J_y^2}$, the angular

506: momentum (per unit mass) in the plane of the disk. The H99 stream

507: stars ({\em large circles}) clearly occupy a different region of phase

508: space than the rest of the halo stars, attesting to the debris' common

509: origin as a Galactic satellite. Figure~\ref{ang_mom_h99} ({\em top

510: panels}) shows cylindrical velocity coordinates.  The stream stars

511: ({\em large circles}) are located in two clumps, one with $v_z \sim

512: -200$~km~s$^{-1}$ and the other with $v_z \sim 200$~km~s$^{-1}$. The

513: bimodality in $v_z$ seen for the nearby stream stars comes about

514: because their orbits reach to a large distance (13 kpc) above and

515: below the Galactic plane. Therefore, any nearby substructure stars

516: will be passing through with high positive or negative $v_z$

517: values. We can anticipate that this bimodality will be less marked as

518: more distant stars are included in the sample, because it will include

519: stream stars with smaller values of $|v_z|$.

520: 

521: %% We did not include the two stars with $J_{\perp} \sim

522: %% 1600$ kpc km/s in the stream because they do not have velocities that

523: %% are consistent with the rest of the stream stars.

524: 

525: Note that the $v_z$ components of the H99 streams extend well into the

526: tails of the velocity distribution, as can be seen in

527: Figure~\ref{vz_hist_h99}.  A smooth halo velocity distribution is

528: well-approximated as a multi-dimensional normal (Gaussian)

529: distribution \citep[e.g.][]{bm98}, so testing the $v_z$ distribution

530: for normality may reveal the presence of substructure like that in

531: H99. In addition, the line of sight component of velocity for stars

532: near the Galactic poles is dominated by the $v_z$ component of the

533: star's velocity.  Therefore, if one uses a sample of stars near the

534: Galactic poles, one should be able to test for structure in the $v_z$

535: distribution using only radial velocities. While with this technique

536: we are only able to identify the presence of substructure in the data,

537: not necessarily the substructure itself, it will be useful in

538: identifying data sets where other observers might want to aggressively

539: pursue proper motions and distances.\footnote{One might imagine

540: employing a genetic-algorithm technique to find stream stars where

541: random groups of stars are selected, the Shapiro-Wilk product

542: calculated, and the groups that have the highest p-values are bred

543: together to create the next generation of groups to test. This is, of

544: course, a computationally intensive method.}

545: 

546: There are many tests for deviations from normality; one of the most

547: powerful is the Shapiro-Wilk test \citep{sw65,da86}.  This test is

548: based on the concept of probability plots, which plot the cumulative

549: distribution function\footnote{The cumulative distribution function

550: gives the fraction of data points that are less than or equal to a

551: value as a function of value: $CDF(x) = P(X \le x)$, where $x$ is the

552: value and $X$ is the data.} using a transformation of the vertical

553: axis that makes normally distributed data fall along a straight line.

554: Examples of probability plots are shown in Chapter 1 of

555: \citet{daste86}.  The slope of this line gives an estimate of $\sigma$

556: for a normal distribution.  The Shapiro-Wilk test compares this slope

557: with the sample standard deviation \citep{st96}. This technique is

558: particularly sensitive to deviations from normality in the tails of

559: the distribution. Code for this test \citep{ro95} can be obtained at

560: StatLib.\footnote{http://lib.stat.cmu.edu/apstat/R94}

561: 

562: %\subsubsection{H99 Data}

563: 

564: To test this method, we applied the Shapiro-Wilk test to the H99 $v_z$

565: data. Here the null hypothesis is that the data have a normal $v_z$

566: distribution, and the probability value (p-value for short) indicates

567: how often data with a normal distribution would produce this

568: data set. Thus the smaller the p-value, the less likely it is that the

569: data come from a normal distribution.  (Note, however, that a high

570: p-value does not mean that the sample distribution is normal, just

571: that it is consistent with a normal distribution.) The results of this

572: test are given in Table~\ref{sw_h99}. We list the tested data set and

573: the associated p-value. The entire sample fails the Shapiro-Wilk test

574: at the 1\% level. When the H99 stream stars and one other star (HD

575: 124358) at ($J_z \sim -1500$, $J_\perp \sim 2300$) are excluded,

576: however, the sample passes the Shapiro-Wilk test. The additional

577: excluded star, indicated by a large triangle, has a very retrograde

578: orbit and is well away from the rest of the data in angular momentum

579: space; it may belong to another stream (see

580: Section~\ref{streams_in_combined}). These results demonstrate the

581: effectiveness of this method for detecting streams like those in H99,

582: where the stars occupy the tails of one component of the velocity

583: distribution.

584: 

585: \subsection{Modeling Stream Detections with Distance} \label{modeling}

586: 

587: Having established that we are able to detect substructure using the

588: $v_z$ velocity distribution, now we use the H99 model of the debris

589: and our model of the smooth halo to see (1) how changing the sample

590: distance limit affects our ability to detect streams and (2) how close

591: stars need to be to the Galactic pole to detect substructure using

592: radial velocity information only.

593: 

594: To characterize stream detections, we produced random samples with

595: sizes ranging from 20 to 4000 stars, and distance limits of 1, 2, 3, 4

596: and 5 kpc.  In each case, we produced 10,000 different realizations of

597: the sample.  While the stream stars are spatially well-mixed up to 5

598: kpc from the Sun, the relative number of stream stars to smooth halo

599: stars drops slowly as we move away from the Sun; see

600: Figure~\ref{frac_stream_plot}. The number of stream stars in each

601: distance-limited sample was scaled using the models so that stream

602: stars made up 10\% of the sample for a distance limit of 1.0~kpc. The

603: maximum distance limit in our simulations was 5~kpc, since the

604: velocity ellipsoid of the halo may change beyond this distance

605: \citep[e.g.][]{vedel}. For each sample, we tested the $v_z$

606: distribution for normality using the Shapiro-Wilk test. We calculated

607: the p-value (significance level) for each of the 10,000 realizations

608: of the sample, and then calculated the average p-value for each

609: sample.

610: 

611: Figure~\ref{n_pw_dist_plot} shows the average p-value as a function of

612: sample size for various distance limits in our $v_z$ samples. P-values

613: below 5\% fail the Shapiro-Wilk test for normality. The number of

614: stars needed to detect the stream with a 1.0~kpc distance limit is

615: less than the sample size of H99 sample with this distance limit (101

616: stars), so it is unsurprising that we were able to detect the H99

617: streams using this method (see Table

618: \ref{meta_fehm10_d25_vz_results}). As the distance limit increases,

619: more stars are needed in the sample to detect the presence of the H99

620: streams. For example, with a distance limit of 5~kpc, approximately

621: 600 stars are needed to detect the H99 streams, but only 75 stars are

622: needed with a distance limit of 1.0 kpc. There are two causes

623: here. First, the fraction of stream stars in a sample decreases slowly

624: as the sample distance limit increases (see

625: Fig.~\ref{frac_stream_plot}). Second, the velocity distribution of

626: the stream changes away from the solar neighborhood, and this results

627: in fewer stream stars with extreme values of $v_z$. (Since the

628: high-energy debris from the progenitor is bound to the Galaxy, stars

629: nearest the plane will have the largest $v_z$ velocities. As the

630: distance from the plane increases, the $v_z$ velocities of the stars

631: will decrease.)  The result of both effects means that increasing the

632: distance limit on a sample does not necessarily increase the detection

633: probability, if it does not have enough stars for the stream to be

634: detected in the sample with the larger distance limits.

635: 

636: To use radial velocities 

637: %(corrected to a galactocentric reference frame) 

638: to identify kinematic structure in $v_z$, we need to identify how

639: close the sample stars need to be to the Galactic poles to have enough

640: $v_z$ velocity information to detect structure. To examine this, we

641: generated five different distance limited samples ($D \le 1.0$, $D \le

642: 2.0$, $D \le 3.0$, $D \le 4.0$, and $D \le 5.0$ kpc) and applied four

643: different Galactic latitude limits to each sample ($|b| \ge 40$, $|b|

644: \ge 50$, $|b| \ge 60$, $|b| \ge 70$, and $|b| \ge 80$).  The stream

645: stars were assumed to be well mixed spatially so we used the same

646: normalization for the stream stars as in the previous set of

647: simulations. We calculated the radial velocity of each star in each

648: sample and then used the same algorithm as above substituting the

649: radial velocities for the $v_z$ velocities to determine an average

650: p-value for the radial velocity distribution of a sample. The results

651: of these simulations are given in Figure~\ref{n_pave_dist_b}.

652: 

653: Comparing Figures~\ref{n_pw_dist_plot} and~\ref{n_pave_dist_b}, one

654: sees that for lower Galactic latitude limits, much larger sample sizes

655: are needed to detect streams. As the Galactic latitude limits get

656: further from the Galactic pole, the width of the smooth halo radial

657: velocity distribution increases, drowning out the presence of stream

658: stars in the wings of the distribution. As we sample closer to the

659: Galactic poles, the sample size needed to detect the H99 streams using

660: radial velocities approaches that needed to detect the streams in $v_z$

661: velocity space.  In short, using the radial velocity test to

662: overcome a lack of proper motion data requires a significant increase

663: in the sample size.

664: 

665: \subsection{Streams in the Combined Data Set} \label{streams_in_combined}

666: 

667: We used the combined sample described in

668: Section~\ref{combined_sample_sec} to investigate further the

669: applicability of the Shapiro-Wilk test to real data.  We concentrate

670: here on looking for outliers in the $J_z$ versus $J_\perp$

671: distribution. In future work (M07),

672: we will investigate structure within the $J_z$ versus $J_\perp$

673: distribution, i.e., within the angular momentum distribution of the

674: disk and the halo themselves, such as in \citet{nhf04}.

675: 

676: While H99 used two different distance limits in their analysis (1 and

677: 2.5 kpc) in order to limit the effect of larger errors in tangential

678: velocity, our direct calculation of errors on the angular momenta make

679: this unnecessary. H99 also limited their sample to stars with $\feh

680: \leq -1.6$ to exclude thick disk stars. We chose simply to exclude

681: such stars using their position on the angular momentum plot. To be

682: considered a thick disk star, the star needed to have a $J_z$ between

683: 1500 and 2500~kpc~km~s$^{-1}$ and a $J_\perp$ less than 600

684: kpc~km~s$^{-1}$. These limits were chosen to match the angular

685: momentum distribution of stars in the \citet{nordstrom04} sample with

686: $\feh$ greater than --1.0 and heliocentric distances less than 2.5

687: kpc. In addition, the star's $v_\phi$ needed to be near 220 $\kms$

688: with all other velocity components near zero. Note that a star

689: traveling at the speed ($220 \ \kms$) and position ($8.0 \ \rm{kpc}$)

690: of the LSR has a $J_z$ of 1760~kpc~km~s$^{-1}$. The $J_\perp$ limit

691: reflects the velocity dispersion of stars in the disk.  These criteria

692: excluded 24 stars, of which seven had $\feh < -1.6$. Note the symmetry

693: about $J_z \sim 0$ for the resulting halo distribution

694: (Fig.~\ref{meta_fehm10_d25_ang_mom_groups}). The clumpiness within the

695: distribution of halo stars will be investigated in M07.

696: 

697: We have identified the H99 stream stars and two other groups of

698: outliers in the combined sample. The properties of the H99 stream

699: stars and the outliers are given in Tables~\ref{meta_strm_table},

700: \ref{meta_out1_table}, and

701: \ref{meta_out2_table}. Figure~\ref{meta_fehm10_d25_ang_mom_groups}

702: shows where these stars fall on the angular momentum and cylindrical

703: velocity plot. The H99 streams are clumped in angular momentum and

704: cylindrical velocity space, but not in Galactic coordinates. The first

705: group of outliers, on the retrograde (left) side of the angular

706: momentum plot, might also be tidal debris. It has extent and isolation

707: similar to the H99 streams. It also forms a kinematically distinct

708: group in the cylindrical velocity plots (at $\avevphi \sim -300$ km

709: s$^{-1}$) as well as being completely made up of low metallicity stars

710: (see Table~\ref{meta_out1_table}; \citet{venn04} have already noted

711: the chemical homogeneity of the extreme retrograde stars in the solar

712: neighborhood). It is not, however, as tightly clumped in velocity

713: space as the H99 streams. The second group of outliers is located on

714: the prograde (lower right) side of the angular momentum plot and has

715: kinematics similar to disk stars, but with higher J$_z$ ($\avevphi

716: \sim 300$ km s$^{-1}$, $\avevR \sim 0$ km s$^{-1}$, $\avevz \sim 0$ km

717: s$^{-1}$).  The prograde outlier group is also very metal poor (see

718: Table~\ref{meta_out2_table}). Because of the closeness to the disk and

719: thick disk region of the angular momentum plot, it is possible that

720: this group is related to the disk or was accreted into the disk by

721: dynamical friction as in~\citet{abadi03}.

722: 

723: We can estimate the probability of stars from a smooth halo populating

724: the above regions in angular momentum space using the smooth halo

725: model described in Section~\ref{model}. We randomly selected 231 stars

726: within 2.5~kpc of the Sun from this model and counted the number of

727: stars in each region of the angular momentum plot where we see

728: outliers in the data. The size of the angular momentum region was

729: chosen to be large enough to enclose all the outliers and their

730: errors. We did not include AS Cnc in the H99 streams region because we

731: are not sure if it is really part of this group. (See

732: Section~\ref{h99_stream_prop}.) We also randomly generated errors for

733: the points from the model based on the distribution of errors in the

734: combined data set. The errors were modeled as normal distributions

735: with mean zero, $\sigma(J_z)= 186 \ \rm{kpc \ km \ s^{-1}}$, and

736: $\sigma(J_\perp) = 107 \ \rm{kpc \ km \ s^{-1}}$. This process was

737: repeated for 100,000 trials. Table~\ref{box_prob} gives the box

738: parameters, the probability that there would be as many or more stars

739: as there are possible stream stars in the box if the halo was entirely

740: smooth, and the average number of stars in the each region for a

741: smooth halo.

742: 

743: For the H99 stream, we found that there was a very low probability of

744: having 11 smooth halo stars in that region of the angular momentum

745: diagram. This result supports the conclusions of H99. In the

746: retrograde outlier region, there is a 7\% chance of having six or more

747: stars in that region of the angular momentum diagram. This set of

748: outliers is possibly another stream.  The simulations tell a different

749: story for the prograde outliers. The probability for having at least

750: three stars in this region of the angular momentum diagram is 70\%. In

751: addition, changing the criterion used to exclude thick disk stars

752: could eliminate the prograde group or fill in the region between the

753: halo distribution and the prograde group. The prograde outliers are

754: likely smooth halo stars (perhaps with some contribution from the

755: thick disk as well) rather than part of a stream.

756: 

757: \subsubsection{H99 Stream Properties} \label{h99_stream_prop}

758: 

759: The H99 streams make up 5\% of our total sample. For a subsample of

760: the combined data set with the same parameters as the H99 sample

761: ($\feh \leq -1.0$ and distances less than 1.0 kpc), the fraction is

762: 9\%. We detect 11 of 12 stars that H99 detected and possibly add one

763: more star to the streams (AS Cnc). The missing H99 star (HD 214925) we

764: group with the retrograde outlier group rather than the H99

765: streams. In Figure~\ref{meta_fehm10_d25_ang_mom_groups}, AS Cnc is

766: located very far from the rest of the H99 group in angular momentum

767: space (at $J_z \sim 4000$~kpc~km~s$^{-1}$ and $J_\perp \sim

768: 3600$~kpc~km~s$^{-1}$) , but agrees with the rest of the stars in

769: velocity space. Therefore, we cannot say for certain whether AS Cnc is

770: part of the H99 group without better data on its distance and space

771: velocities. We include AS Cnc in our list of members of the H99

772: streams for completeness, not because we are certain it is a member.

773: 

774: \citet{katie} and \citet{2000AJ....120.1841F,fulbright2002} gave

775: high-dispersion abundance analyses for three of the H99 stream stars

776: in Table 4: HD 128279, BD+30 2611 and HD 175305. Their [Mg/Fe] and

777: [$\alpha$/Fe] abundances are similar to other halo stars in the solar

778: neighborhood, suggesting that the gas that formed these stars was

779: enriched mostly by Type II supernovae.

780: 

781: \citet{fiorentin05} has performed a similar analysis to ours using

782: only the B00 data and finds seven additional stars in the H99

783: streams. Two of the \citet{fiorentin05} stars (HD 214161 and BPS CS

784: 22189-0007) we excluded from our sample because they were classified

785: by B00 as giants and had $(B-V)_0$ less than 0.9; they are likely to

786: have underestimated distances (see

787: Section~\ref{combined_rg}). \citet{fiorentin05} also include RZ Cep in

788: their list of members of the H99 streams. RZ Cep is a type c RR Lyrae

789: and thus not included in the M07 sample of RR Lyrae stars; it could be

790: an additional member of the stream.  Finally, five of the stars that

791: \citet{fiorentin05} detect (BPS CS 22948-0093, BPS CS 30339-0037, BPS

792: CS 29513-0031, BPS CS 29504-0044, and BPS CS 22876-0040) are

793: classified by B00 as turnoff stars. We did not include these stars in

794: our sample because the distance estimates used in B00 for turnoff

795: stars were based on UBV-photometry and thus not able to deal with the

796: evolution of turnoff stars up to the subgiant branch. This ambiguity

797: introduces additional uncertainty into the distance estimates for

798: these stars \citep[][especially their Fig. 8]{schuster04}.

799: 

800: We tested the $v_z$ velocity data for normality using the Shapiro-Wilk

801: test. The results of these tests are given in

802: Table~\ref{meta_fehm10_d25_vz_results}. When the H99 stream stars are

803: excluded from the sample, it tests positive for normality. Comparing

804: our results to the simulations in Section~\ref{modeling}, it is not

805: surprising that we were able to detect the H99 streams in this sample

806: using only $v_z$ velocities. We did not test the radial velocity

807: distribution of this sample for normality because there are not enough

808: stars in the sample for this test to detect structure.

809: 

810: We can use the observed asymmetry in the $v_z$ velocity distribution

811: of the H99 streams along with our model of them to estimate the how

812: long ago the progenitor was accreted. Initially, all stars are bound,

813: and as time goes by, the stars are released leading to the formation

814: of streams.  These streams will phase-mix, and after many gigayears,

815: this mixing will have progressed so that half the stars will cross the

816: Galactic disk in each direction. In

817: Figure~\ref{meta_fehm10_d25_ang_mom_groups}, we see that the observed

818: fraction of stars in the stream with negative $v_z$ is 0.72 (8/11).

819: To estimate the age of the H99 streams, we selected 11 model stars

820: within 2.5 kpc of the Sun from our simulation and determined the

821: fraction in the stream with negative $v_z$. We repeated this 1000

822: times to build up a probability distribution. We then compared the

823: observed fraction of stars ($8/11=0.72$) to this probability

824: distribution to determine an age (Fig.~\ref{model_vz_evolution}). We

825: find that the observed fraction is matched by the mean simulated

826: fraction for a progenitor that was accreted between 6 and 9~Gyr

827: ago. Although we cannot determine a definite age, we can rule out the

828: possibility that the stream is either very young (3~Gyr) or very old

829: (12~Gyr). Future observations can improve this situation. By

830: increasing the observed total number of stars in the streams, the

831: width of the peak of the probability distribution in

832: Figure~\ref{model_vz_evolution} decreases allowing a more accurate

833: determination of the age of the accretion event.

834: 

835: %% We can use the observed asymmetry in the $v_z$ velocity distribution

836: %% of the H99 streams along with our model of them to determine how long

837: %% ago the progenitor was disrupted. At accretion, all the model stars

838: %% are in one stream, and after 12 Gyr the mixing has progressed so that

839: %% half the stars are going in each direction. In

840: %% Figure~\ref{meta_fehm10_d25_ang_mom_groups}, we see that the observed

841: %% fraction of stars in the initial stream is 0.72 (8/11). Note that we

842: %% assume that the initial stream is the stream with more members because

843: %% we expect the fraction of stars in the initial stream to be between

844: %% 0.5 and 1.0. To estimate the age of the H99 streams, we selected 11

845: %% stars from our model stream distribution and determined the fraction

846: %% of the total number of stars in the initial stream. We repeated this

847: %% 10,000 times to build up a probability distribution. We then compared

848: %% the observed fraction of stars in the initial stream ($8/11=0.72$) to

849: %% the probability distribution to determine an age

850: %% (Figure~\ref{model_vz_evolution}). We find that the observed fraction

851: %% of stars in the initial stream matches the mean simulated fraction of

852: %% stars in the initial stream for a progenitor that was disrupted

853: %% between 6 and 9~Gyr ago. Although we cannot determine a definite age,

854: %% we can rule out the possibility that the stream is either very young

855: %% (3~Gyr) or very old (12~Gyr). Future observations can improve this

856: %% situation. By increasing the observed total number of stars in the

857: %% streams, the width of the peak of the probability distribution in

858: %% Figure~\ref{model_vz_evolution} decreases allowing a more accurate

859: %% determination of the age of the stellar stream.

860: 

861: 

862: %% We find that the observed fraction

863: %% of stars in the initial stream matches the mean simulated fraction of

864: %% stars in the initial stream for a progenitor that was disrupted

865: %% approximately 6~Gyr ago. However, there is a significant probability

866: %% that the stream could be as old at 9 Gyr.

867: 

868: %% Old paragraph with slight different take on the situation.

869: 

870: %% We can use the observed asymmetry in the $v_z$ velocity distribution

871: %% of the H99 streams along with our model of them to determine how long

872: %% ago the progenitor was disrupted. At accretion, there are no model

873: %% stars with negative $v_z$, and after 12 Gyr the mixing has progressed

874: %% so that half the stars are going in each direction. We selected 11

875: %% stars from our model stream distribution and determined the fraction

876: %% of the total number of stars in the negative $v_z$ stream as a

877: %% function of the time since accretion. We repeated this 1,000 times to

878: %% build up a probability distribution. We then compared the observed

879: %% fraction of stars with negative $v_z$ velocities ($3/11=0.27$) to the

880: %% probability distribution (Figure~\ref{model_vz_evolution}). We find

881: %% that the observed fraction of stars in the negative $v_z$ stream

882: %% matches the simulated mean fraction of stars in this stream for a

883: %% progenitor that was disrupted approximately 6~Gyr ago.

884: 

885: 

886: \subsubsection{Retrograde Outlier Properties}

887: 

888: The retrograde outliers do not add significant structure to the $v_z$

889: velocity distribution (the sample passes the Shapiro-Wilk test if the

890: H99 streams, but not the retrograde outliers, are removed), but they

891: do have significant structure in the $v_\phi$ direction, as can be

892: seen in Figure~\ref{meta_fehm10_d25_ang_mom_groups}. To see if the

893: Shapiro-Wilk test is able to pick out these deviations from normality,

894: we ran various samples of $v_\phi$ data through the Shapiro-Wilk

895: test. The results are shown in

896: Table~\ref{meta_fehm10_d25_vphi_results}. Excluding the retrograde

897: group of stars (Fig. ~\ref{meta_fehm10_d25_ang_mom_groups}, {\em

898: inverted triangles}) causes the sample to pass the Shapiro-Wilk test

899: (test positive for normality). Indeed, all the samples that pass the

900: Shapiro-Wilk test exclude this group of stars, which are very far from

901: the rest of the $v_\phi$ velocities. Samples excluding only the

902: prograde outliers do not test positive for normality. Evidently

903: removing the prograde outliers from the sample just removes stars from

904: the tail of the normal distribution of the halo and does not remove

905: substructure.

906: 

907: We note that this group of stars has an extremely low value of

908: $J_z$. The 38 globular clusters with proper motion estimates

909: summarized by \citet{dinescu}, most of which are within 10 kpc of the

910: Galactic center, have $J_z$ values ranging from --664 to 2307

911: kpc~$\kms$, while the mean value of $J_z$ for our retrograde group is

912: --2500 kpc~$\kms$.  However, larger samples of nearby proper-motion

913: selected stars have produced stars with even more extreme negative

914: velocities, for example Kapteyn's star group, which has a mean $J_z$

915: around --4000 kpc~$\kms$ \citep{eggen96}.

916: 

917: Three of the stars from the retrograde substructure also have

918: high-dispersion abundance information from \citet{katie},

919: \citet{1995AJ....109.2757M}, and \citet{1988A&A...204..193G}: HD 6755,

920: HD 200654 and CD--24 1782. Interestingly, these stars show a range of

921: [$\alpha$/Fe]: one has roughly normal [$\alpha$/Fe] for the local halo

922: while the other two have low [$\alpha$/Fe] for their [Fe/H]. Note that

923: our retrograde group corresponds roughly to the ``extreme retrograde''

924: class of \citet{venn04}, and we find a similar behavior (somewhat

925: lower [$\alpha$/Fe] than normal) with our improved kinematical

926: measures.

927: 

928: \subsubsection{Discussion}

929: 

930: Including both the H99 streams and the retrograde outliers gives a

931: substructure fraction of 7\% for the halo.  The fraction of stream

932: stars in the local halo is therefore at least 5\% and possibly as high

933: as 7\% in this sample. \citet{gould03} produced an estimate of the

934: overall amount of substructure in the local halo (its ``granularity'')

935: using an updated version on the NLTT proper motion sample. This method

936: uses the fact that dominant streams in the solar neighborhood will not

937: show spatial substructure over such a small volume, but will produce

938: correlations between different components of velocity (non-zero cross

939: terms in the velocity dispersion tensor). These will be reflected in

940: the proper motion distribution. The fraction of stream stars in our

941: sample is consistent with the upper limit set in \citet{gould03}, that

942: one stream can comprise no more than 5\% of the halo.

943: 

944: Chiba and Beers (2000) noted that when they increase the sample size of

945: halo stars to 3 times that of H99 (728 stars), they only find nine

946: stream stars total. There are several reasons why they may not have

947: found more stream stars. First, the box that they use in their

948: Figure~15 to identify the stream is too small. Our

949: Figure~\ref{ang_mom_am} shows the distribution of stars in the model

950: within 2.5~kpc in angular momentum and velocity space. This

951: distribution is much larger than the box that Chiba and Beers (2000)

952: selects.  They also do not remove stars with thick disk kinematics

953: ($J_z \sim 1750$ kpc km$^{-1}$ and $J_\perp \sim 0$ kpc km$^{-1}$)

954: from their sample, thus decreasing the percentage of H99 stream stars.

955: Note, however, that the clumps in the $v_\phi-v_z$ velocity space are

956: still apparent in their diagram, so it is still possible to pick out

957: the H99 streams in their larger sample. This example illustrates the

958: need to look not only at a angular momentum plot, but also a velocity

959: plot to confirm membership in a moving group.

960: 

961: \section{Conclusions} \label{conclusions}

962: 

963: We have assembled a sample of halo stars in the solar neighborhood to

964: look for substructure in velocity and angular momentum space. Our

965: sample of 231 stars includes red giants, RR Lyrae variable stars, and

966: RHB stars within 2.5 kpc of the Sun with [Fe/H] less than --1.0. It

967: was chosen to include stars with well-quantified errors and accurate

968: distances, space velocities, and metallicities. Understanding the

969: errors in the measured and derived quantities, especially distance, is

970: crucial for this work since they may distort any underlying

971: substructure.

972: 

973: %% Thick disk stars were removed based on their angular momentum. Errors

974: %% for 214 out of 234 stars were determined using Monte Carlo

975: %% simulations.

976: 

977: With our data set, we confirm the existence of the streams found by

978: H99, which we refer to as the H99 streams. These streams have

979: significant structure in their velocity distribution in the $z$

980: direction (out of the Galactic plane). We use the results of H99 to

981: test how one might use $v_z$ velocity information and radial velocity

982: information to detect kinematic substructure in the halo. We find that

983: detecting the H99 streams with radial velocities alone would require a

984: large sample (e.g., approximately 150 stars within 2 kpc of the Sun

985: and within 20$^\circ$ of the Galactic poles). We also use the

986: structure in the velocity distribution of the H99 streams to estimate

987: the age of this group. From our model of the H99 progenitor, we

988: determine that the H99 streams' progenitor was accreted between 6 and

989: 9~Gyr ago.

990: 

991: We have also discovered, in angular momentum space, two other possible

992: substructures, which we refer to as the retrograde and prograde

993: outliers. For the retrograde outliers, there is a low probability of

994: that region of the angular momentum diagram being occupied by six or

995: more smooth halo stars. Based on this evidence, the retrograde

996: outliers are likely members of a stream. The prograde outliers,

997: however, are most likely smooth halo stars (perhaps with some

998: contribution from the thick disk as well) rather than part of a

999: stream.  The retrograde outliers display significant structure in the

1000: $v_\phi$ direction. Samples excluding the retrograde outliers pass our

1001: test for normality in the $v_\phi$ direction. The fraction of stars in

1002: our sample that are stream stars is between 5\% and 7\%.

1003: 

1004: For H99 streams, the [Mg/Fe] and [$\alpha$/Fe] abundances are similar

1005: to other halo stars in the solar neighborhood, suggesting that the gas

1006: that formed these stars was enriched mostly by Type II

1007: supernovae. The retrograde outliers show a range of [$\alpha$/Fe]: one

1008: has roughly normal [$\alpha$/Fe] for the local halo while the other

1009: two have low [$\alpha$/Fe] for their [Fe/H]. Note that our retrograde

1010: group corresponds roughly to the ``extreme retrograde'' class of

1011: \citet{venn04}, and we find a similar behavior (somewhat lower

1012: [$\alpha$/Fe] than normal) with our improved kinematical measures.

1013: 

1014: Although we are not the first to note that stellar debris from the

1015: disruption of a satellite would have a double-peaked distribution in

1016: galactocentric radial velocity \citep[see also][]{meza05}, the methods

1017: developed in this paper add to the toolbox of kinematic methods

1018: \citep[e.g. H99][]{h06} being developed to exploit future large

1019: databases such as RAVE \citep{rave2006}, SDSSII/SEGUE

1020: \citep{segue_paper}, and {\em Gaia} \citep{gaiabook} to detect kinematic

1021: substructure in our Galaxy's halo. These tools, in conjunction with

1022: studies of spatial over-densities in the Milky Way

1023: \citep[e.g.][]{willman02,belokurov2006}, will provide crucial answers

1024: to the puzzle of how our Galaxy was formed.

1025: 

1026: \acknowledgments

1027: 

1028: A. A. K. was supported by a NSF Graduate Research Fellowship during

1029: portions of this work. H. L. M. acknowledges the support of NSF grant

1030: AST-0098435. A. A. K. would like to thank Eric M. Wilcots for his

1031: patience while she finished this paper, and H. L. M. thanks Bruce

1032: Twarog and Barbara Anthony-Twarog for their helpful explanations of

1033: the $uvby$ luminosity classifications and Kim Venn for a useful

1034: discussion on [$\alpha$/Fe] measurements. The authors would also like

1035: to thank the referee for his or her helpful comments.  We made

1036: extensive use of the SIMBAD astronomical database for this project.

1037: 

1038: \appendix

1039: 

1040: \section{Transforming Local Heliocentric Coordinates to Galactocentric

1041: Model Coordinates} \label{transform}

1042: 

1043: Given the wide variety of coordinate systems used in Galactic

1044: astronomy, we explicitly derive the transformation between the local,

1045: Sun-centered coordinate systems $(x',y',z')$ used in this paper and

1046: the Galactocentric coordinate system used in the H99 models

1047: $(x,y,z)$. Figure~\ref{gal_coords} illustrates the two coordinate

1048: systems. The local velocity coordinate system is a left-handed

1049: coordinate system with the $x'$ axis pointing away from the Galactic

1050: center, the $y'$ axis pointing in the direction of Galactic rotation,

1051: and the $z'$ axis pointing in the direction of the NGP. Note that the

1052: $L_{z'}$ angular momentum vector points in the direction opposite of

1053: the $z'$ axis. In other words, the Galaxy rotates clockwise when you

1054: look down from the NGP. The model coordinate system used in H99 is

1055: also left-handed.

1056: 

1057: The position of a star with respect to the Sun is given by 

1058: \begin{eqnarray}

1059: x' & = & - d \cos (b) \cos (l) \\

1060: y' & = & d \cos (b) \sin (l) \\

1061: z' & = & d \sin (b)

1062: \end{eqnarray}

1063: where $d$ is the distance to the star in kpc, $b$ is the Galactic latitude,

1064: and $l$ is the Galactic longitude. The Galactocentric model

1065: coordinates are then

1066: \begin{eqnarray}

1067: x & = & 8.0 \ \mathrm{kpc} + x'  \\

1068: y & = & y' \\

1069: z & = & z'

1070: \end{eqnarray}

1071: where 8.0~kpc is the distance between the Sun and the Galactic

1072: center. 

1073: 

1074: To transform the observed space velocities to the Local

1075: Standard of Rest (LSR) frame,

1076: we use the equations

1077: \begin{eqnarray}

1078: v_{x'} & = & U + u_\odot \\

1079: v_{y'} & = & V + v_\odot + v_{lsr} \\

1080: v_{z'} & = & W + w_\odot 

1081: \end{eqnarray}

1082: where $U$, $V$, and $W$ are the space velocities directed toward the

1083: Galactic anti-center, toward the direction of rotation, and toward the

1084: NGP. We used the values $v_{lsr} = 220.0 \ \kms$,

1085: $u_{\odot} = -9.0 \ \kms$, $v_{\odot} = 12.0 \ \kms$, and $w_{\odot} =

1086: 7.0 \ \kms$ \citep{blaauw,mb81} to correct for the motions of the Sun

1087: and the LSR. These velocities are the velocities in the model frame, i.e.,

1088: the following

1089: \begin{eqnarray}

1090: v_x & = & v_{x'}   \\

1091: v_y & = & v_{y'}  \\

1092: v_z & = & v_{z'}.

1093: \end{eqnarray}

1094: 

1095: The angular momentum components (per unit mass) are then the

1096: cross products

1097: \begin{eqnarray}

1098: J_x & = & y v_z - v_y z \\

1099: J_y & = & z v_x - v_z x \\

1100: J_z & = & x v_y - v_x y

1101: \end{eqnarray}

1102: and $J_\perp$ is $\sqrt{ J_x^2 + J_y^2}$. Note that these angular

1103: momentum products are calculated on a left-handed system. While this

1104: does not make a difference in $J_\perp$, a left-handed $J_z$ points in

1105: the opposite direction as a right-handed $J_z$. In other words, the

1106: left-handed $J_z$ of a star near the Sun is $220 \ \kms \times 8 \

1107: \mathrm{kpc} = 1760 \ \mathrm{kpc} \ \kms$, while the traditional

1108: right-handed $J_z$ of a star near the Sun is $-1760 \ \mathrm{kpc} \

1109: \kms$.

1110: 

1111: \begin{thebibliography}{}

1112: 

1113: \bibitem[Abadi et al.(2003)]{abadi03} Abadi, M.~G., Navarro, 

1114: J.~F., Steinmetz, M., \& Eke, V.~R.\ 2003, \apj, 597, 21 

1115: 

1116: \bibitem[Anthony-Twarog \& Twarog(1994)]{att94}Anthony-Twarog, B.~J. \&

1117: Twarog, B.~A. 1994, \aj, 107, 1577

1118: 

1119: %%\bibitem[Beers et al.(1990)]{b90}Beers, T., Preston, G., Shectman,

1120: %%S., \& Kage, J. 1990, \aj, 100, 849

1121: 

1122: \bibitem[Beers et al.(2004)]{segue_paper} Beers, T.~C., Allende

1123: Prieto, C., Wilhelm, R., Yanny, B., \& Newberg, H.\ 2004, Publ.

1124: Astron. Soc. Australia, 21, 207

1125: 

1126: \bibitem[Beers et al.(2000)]{b00}Beers, T.~C., Chiba, M., Yoshii, Y.,

1127: Platais, I., Hanson, R.~B., Fuchs, B., \& Rossi, S. 2000, \aj, 119, 2866

1128: 

1129: \bibitem[Beers \& Sommer-Larsen(1995)]{bs95}Beers, T.~C. \&

1130: Sommer-Larsen, J. 1995, \apjs, 96, 175

1131: 

1132: %%\bibitem[Beers et al.(2002)]{b02} Beers, T.~C., Drilling, J.~S.,

1133: %%Rossi, S., Chiba, M., Rhee, J., F{\" u}hrmeister, B., Norris, J.~E.,

1134: %%\& von Hippel, T.\ 2002, \aj, 124, 931

1135: 

1136: \bibitem[Belokurov et al.(2006)]{belokurov2006} Belokurov, V., et 

1137: al.\ 2006, \apjl, 642, L137 

1138: 

1139: %%\bibitem[Bidelman \& MacConnell(1973)]{bm73}Bidelman, W. \&

1140: %%MacConnell, D. 1973, \aj, 78, 687

1141: 

1142: \bibitem[Binney \& Merrifield(1998)]{bm98}Binney, J. \& Merrifield,

1143: M. 1998, Galactic Astronomy, (3rd ed.; Princeton, N.J.: Princeton

1144: University Press)

1145: 

1146: \bibitem[Blaauw \& Schmidt(1965)]{blaauw} Blaauw, A. and Schmidt,

1147: M. 1965, {\it Galactic Structure}, (Chicago, University of Chicago

1148: Press).

1149: 

1150: \bibitem[Bond(1980)]{b80}Bond, H.~E. 1980, \apjs, 44, 517

1151: 

1152: \bibitem[Brown et al.(1999)]{brown99} Brown, 

1153: J.~A., Wallerstein, G., \& Gonzalez, G.\ 1999, \aj, 118, 1245 

1154: 

1155: \bibitem[Brown et al.(1997)]{brown97} Brown, 

1156: J.~A., Wallerstein, G., \& Zucker, D.\ 1997, \aj, 114, 180 

1157: 

1158: %%\bibitem[Carney \& Latham(1986)]{c86}Carney, B. \& Latham, D. \aj, 92, 60

1159: 

1160: \bibitem[Carney et al.(1997)]{carney97} Carney, B.~W., Wright, 

1161: J.~S., Sneden, C., Laird, J.~B., Aguilar, L.~A., \& Latham, D.~W.\ 1997, 

1162: \aj, 114, 363 

1163: 

1164: \bibitem[Carney et al.(2003)]{carney03} Carney, B.~W., Latham, 

1165: D.~W., Stefanik, R.~P., Laird, J.~B., \& Morse, J.~A.\ 2003, \aj, 125, 293 

1166: 

1167: \bibitem[Chiba \& Beers(2000)]{cb00}Chiba, M. \& Beers, T.~C. 2000, \aj,

1168: 119, 2843

1169: 

1170: \bibitem[Chiba \& Yoshii(1998)]{cy98}Chiba, M. \& Yoshii, Y. 1998,

1171: \aj, 115, 168

1172: 

1173: \bibitem[D'Agostino(1986)]{da86}D'Agostino, R.~B. 1986, in

1174: Goodness-of-Fit Techniques. ed. R.~B. D'Agostino \& M.~A. Stephens (New

1175: York: Dekker), 367

1176: 

1177: \bibitem[D'Agostino \& Stephens(1986)]{daste86} D'Agostino, R. \&

1178:   Stephens, M., eds. 1986. Goodness-of-Fit Techniques. New York:

1179:   Marcel Dekker, Inc.

1180: 

1181: \bibitem[Dinescu et al.(1999)]{dinescu} Dinescu, D.~I., Girard, 

1182: T.~M., \& van Altena, W.~F.\ 1999, \aj, 117, 1792 

1183: 

1184: %%\bibitem[Dohm-Palmer et al.(2000)]{dp00}Dohm-Palmer, R.; Mateo, M.;

1185: %%Olszewski, E.; Morrison, H.L.; Harding, P.; Freeman, K..; Norris,

1186: %%J. 2000, \aj, 120, 2496

1187: 

1188: \bibitem[Eggen(1996)]{eggen96} Eggen, O.~J.\ 1996, \aj, 112, 

1189: 1595

1190: 

1191: %%\bibitem[Eggen(1998)]{e98}Eggen, O. 1998, \aj, 115, 2397

1192: 

1193: \bibitem[Eggen et al.(1962)]{els62}Eggen, O.~J.,

1194: Lynden-Bell, D., \& Sandage, A.~R. 1962, \apj, 136, 748

1195: 

1196: \bibitem[ESA(1997)]{hipcat} ESA 1997, VizieR Online Data 

1197: Catalog, 1239, 0 

1198: 

1199: %% \bibitem[Fernley et al.(1993)]{fernley93} Fernley, J.~A., 

1200: %% Skillen, I., \& Burki, G.\ 1993, \aaps, 97, 815 

1201: 

1202: \bibitem[Fiorentin et al.(2005)]{fiorentin05} Fiorentin, P.~R., 

1203: Helmi, A., Lattanzi, M.~G., \& Spagna, A.\ 2005, \aap, 439, 551 

1204: 

1205: \bibitem[Freeman \& Bland-Hawthorn(2002)]{kcf_josh} Freeman, K., 

1206: \& Bland-Hawthorn, J.\ 2002, \araa, 40, 487 

1207: 

1208: \bibitem[Fulbright(2000)]{2000AJ....120.1841F} Fulbright, J.~P.\ 2000, \aj, 

1209: 120, 1841

1210:  

1211: \bibitem[Fulbright(2002)]{fulbright2002} Fulbright, J.~P.\ 2002, \aj, 

1212: 123, 404 

1213: 

1214: %%\bibitem[Gilmore \& Reid(1983)]{gr83} Gilmore, G.~\&

1215: %%Reid, N.\ 1983, \mnras, 202, 1025 

1216: 

1217: \bibitem[Gould(2003)]{gould03} Gould, A.\ 2003, \apjl, 592, L63 

1218: 

1219: \bibitem[Gratton \& Sneden(1988)]{1988A&A...204..193G} Gratton, R.~G., \& 

1220: Sneden, C.\ 1988, \aap, 204, 193 

1221: 

1222: %%\bibitem[Hambly et al.(2001)]{ha02}Hambly, N., MacGillivray, H., Read, M., Tritton, S., Thomson, E., Kelly, B., Morgan, D., Smith, R., Driver, S., Williamson, J., Parker, Q., Hawkins, M., Williams, P., \& Lawrence A. 2001, \mnras, 326, 1279

1223: 

1224: %%\bibitem[Harris(1996)]{ha96}Harris, W. 1996, \aj, 112, 1487

1225: 

1226: \bibitem[Hanson et al.(1998)]{hanson} Hanson, R.~B., Sneden, 

1227: C., Kraft, R.~P., \& Fulbright, J.\ 1998, \aj, 116, 1286 

1228: 

1229: \bibitem[Hartkopf \& Yoss(1982)]{hy82} Hartkopf, W.~I., \& 

1230: Yoss, K.~M.\ 1982, \aj, 87, 1679 

1231: 

1232: \bibitem[Helmi et al.(2006)]{h06} Helmi, A., Navarro, 

1233: J.~F., Nordstr{\"o}m, B., Holmberg, J., Abadi, M.~G., \& Steinmetz, M.\ 

1234: 2006, \mnras, 365, 1309 

1235: 

1236: \bibitem[Helmi \& White(1999)]{hw99}Helmi, A. \& White, S.~D.~M. 1999,

1237: \mnras, 307, 495

1238: 

1239: \bibitem[Helmi et al.(1999)]{h99}Helmi, A., White, S.~D.~M., de Zeeuw, P.~T.,

1240: \& Zhao, H. 1999, \nat, 402, 53

1241: 

1242: \bibitem[Ibata et al.(1994)]{igi94}Ibata, R.~A., Gilmore, G.,

1243: \& Irwin, M.~J. 1994, \nat, 370, 194

1244: 

1245: \bibitem[Ivans et al.(2003)]{inese03} Ivans, I.~I., Sneden, C., 

1246: James, C.~R., Preston, G.~W., Fulbright, J.~P., H{\"o}flich, P.~A., Carney, 

1247: B.~W., \& Wheeler, J.~C.\ 2003, \apj, 592, 906 

1248: 

1249: %%\bibitem[Johnson \& Soderblom(1987)]{js87}Johnson, D. \& Soderblom,

1250: %%D. 1987, \aj, 93, 864

1251: 

1252: \bibitem[King(1997)]{king97} King, J.~R.\ 1997, \aj, 113, 2302 

1253: 

1254: %%\bibitem[Kinman(2003)]{ki03}Kinman, T. 2003, in preparation

1255: 

1256: %%\bibitem[Kholopov(1985)]{kh85}Kholopov, P.N. 1985, General Catalogue

1257: %%of Variable Stars (Moskva: Nauka)

1258: 

1259: \bibitem[Layden(1994)]{l94}Layden, A.~C. 1994, \aj, 108, 1016

1260: 

1261: \bibitem[Layden et al.(1996)]{l96}Layden, A.~C., Hanson, R.~B., Hawley,

1262: S.~S., Klemola, A.~R., \& Hanley, C.~J. 1996, \aj, 112, 2110

1263: 

1264: \bibitem[Liu(1991)]{liu} Liu, T.\ 1991, \pasp, 103, 205 

1265: 

1266: \bibitem[Lynden-Bell \& Lynden-Bell(1995)]{lb2} 

1267: Lynden-Bell, D., \& Lynden-Bell, R.~M.\ 1995, \mnras, 275, 429

1268: 

1269: \bibitem[McWilliam et al.(1995)]{1995AJ....109.2757M} McWilliam, A., 

1270: Preston, G.~W., Sneden, C., \& Searle, L.\ 1995, \aj, 109, 2757 

1271: 

1272: \bibitem[Martin \& Morrison(1998)]{mm98}Martin, J.C., \& Morrison,

1273: H.L. 1998, \aj, 116, 1724

1274: 

1275: \bibitem[Meza et al.(2005)]{meza05} Meza, A., Navarro, J.~F., 

1276: Abadi, M.~G., \& Steinmetz, M.\ 2005, \mnras, 359, 93 

1277: 

1278: \bibitem[Mihalas \& Binney(1981)]{mb81}Mihalas, D., \& Binney,

1279: J. 1981, Galactic Astronomy (2nd ed.; San Francisco: W.H. Freeman)

1280: 

1281: \bibitem[Morrison et al.(1990)]{mff90}Morrison, H.L., Flynn,

1282: C., \& Freeman, K.C. 1990, \aj, 100, 1191

1283: 

1284: %%\bibitem[Morrison et al.(2000)]{mo00}Morrison, H.L., Mateo, M.,

1285: %%Olszewski, E., Harding, P., Dohm-Palmer, R., Freeman, K., Norris, J.,

1286: %%Morita, M. 2000, \aj, 119, 2254

1287: 

1288: %%\bibitem[Morrison et al.(2001)]{mo01}Morrison, H.L., Olszewski, E., Mateo, M., Norris, J., Harding, P., Dohm-Palmer, R., Freeman, K. 2001, \aj, 121, 283

1289: 

1290: \bibitem[Morrison et al.(2003)]{mo03}Morrison, H.L., et al. 2003, AJ,

1291: 125, 2502

1292: 

1293: \bibitem[Navarro et al.(2004)]{nhf04} Navarro, J.~F., Helmi, 

1294: A., \& Freeman, K.~C.\ 2004, \apjl, 601, L43 

1295: 

1296: \bibitem[Nissen \& Schuster(1997)]{nissen97} Nissen, P.~E., \& 

1297: Schuster, W.~J.\ 1997, \aap, 326, 751 

1298: 

1299: \bibitem[Nordstr{\"o}m et al.(2004)]{nordstrom04} Nordstr{\"o}m, 

1300: B., et al.\ 2004, \aap, 418, 989 

1301: 

1302: \bibitem[Norris et al.(1985)]{nbp} Norris,

1303: J., Bessell, M.~S., \& Pickles, A.~J.\ 1985, \apjs, 58, 463

1304: 

1305: %% \bibitem[Norris(1986)]{n86}Norris, J. 1986, \apjs, 61, 667

1306: 

1307: %%\bibitem[Ochsenbein, Bauer, \& Marcout(2000)]{obm00}Ochsenbein F.,

1308: %%Bauer P., \& Marcout J., 2000, \aap, 143, 221

1309: 

1310: %%\bibitem[Perryman \& ESA(1997)]{hip}Perryman, M. \& ESA. 1997, The

1311: %%Hipparcos and Tycho catalogues (1st ed.; Noordwijk, Netherlands: ESA

1312: %%Publications)

1313: 

1314: %%\bibitem[Piotto, Cool, \& King(1997)]{pck97}Piotto, G., Cool, A., \&

1315: %%King, I. 1997, \aj, 113, 1345

1316: 

1317: \bibitem[Pilachowski et al.(1996)]{katie} Pilachowski, C.~A., 

1318: Sneden, C., \& Kraft, R.~P.\ 1996, \aj, 111, 1689 

1319: 

1320: \bibitem[Royston(1995)]{ro95}Royston, P. 1995, Applied Statistics, 44,

1321: 547

1322: 

1323: %%\bibitem[Ryan \& Norris(1991)]{ryan91} Ryan, S.~G., \& Norris, 

1324: %%J.~E.\ 1991, \aj, 101, 1865 

1325: 

1326: %%\bibitem[Ryan \& Lambert(1995)]{rl95} Ryan, S.~G.~\& 

1327: %%Lambert, D.~L.\ 1995, \aj, 109, 2068 

1328: 

1329: \bibitem[Schuster et al.(2004)]{schuster04} Schuster, W.~J., 

1330: Beers, T.~C., Michel, R., Nissen, P.~E., \& Garc{\'{\i}}a, G.\ 2004, \aap, 

1331: 422, 527 

1332: 

1333: \bibitem[Schwarzschild(1907)]{swar} Schwarzschild, K. 1907,

1334:   Nachrichten von der K{\"o}niglichen Gesellschaft der Wissenschaften

1335:   zu G{\"o}ttingen, --, 614

1336: 

1337: \bibitem[Searle \& Zinn(1978)]{sz78}Searle, L. \& Zinn, R. 1978, \apj,

1338: 225, 357

1339: 

1340: \bibitem[Shapiro \& Wilk(1965)]{sw65}Shapiro, S.S. \& Wilk, M.B. 1965,

1341:   Biometrika, 52, 591

1342: 

1343: \bibitem[Shetrone et al.(2001)]{shetrone01} Shetrone, M.~D., 

1344: C{\^o}t{\'e}, P., \& Sargent, W.~L.~W.\ 2001, \apj, 548, 592 

1345: 

1346: \bibitem[Shetrone et al.(2003)]{shetrone03} Shetrone, M., Venn, 

1347: K.~A., Tolstoy, E., Primas, F., Hill, V., \& Kaufer, A.\ 2003, \aj, 125, 

1348: 684 

1349: 

1350: \bibitem[Steinmetz \& Muller(1994)]{sm94}Steinmetz, M. \& Muller,

1351: E. 1994, \aap, 281, L97.

1352: 

1353: \bibitem[Steinmetz et al.(2006)]{rave2006} Steinmetz, M., et 

1354: al.\ 2006, \aj, 132, 1645 

1355: 

1356: \bibitem[Stephens \& Boesgaard(2002)]{stephens2002} Stephens, A., \& 

1357: Boesgaard, A.~M.\ 2002, \aj, 123, 1647 

1358: 

1359: 

1360: \bibitem[Stephens(1986)]{st96}Stephens, M.~A. 1986, in Goodness-of-Fit

1361: Technqiues. ed. R.~B. D'Agostino \& M.~A. Stephens (New York: Dekker), 195

1362: 

1363: 

1364: \bibitem[Tolstoy et al.(2003)]{tolstoy03} Tolstoy, E., Venn, 

1365: K.~A., Shetrone, M., Primas, F., Hill, V., Kaufer, A., \& Szeifert, T.\ 

1366: 2003, \aj, 125, 707 

1367: 

1368: \bibitem[Turon et al.(2005)]{gaiabook} Turon, C., O'Flaherty, 

1369: K.~S., \& Perryman, M.~A.~C.\ 2005, ESA SP-576: The Three-Dimensional 

1370: Universe with Gaia

1371: 

1372: \bibitem[Twarog \& Anthony-Twarog(1994)]{tat94} Twarog, 

1373: B.~A.~\& Anthony-Twarog, B.~J.\ 1994, \aj, 107, 1371 

1374: 

1375: \bibitem[Urban et al.(1998)]{ucw98}Urban, S.~E., Corbin, T.~E., \&

1376: Wycoff, G.~L. 1998, \aj, 115, 2161

1377: 

1378: \bibitem[Vedel \& Sommer-Larsen(1990)]{vedel} Vedel, H., \& 

1379: Sommer-Larsen, J.\ 1990, \apj, 359, 104 

1380: 

1381: \bibitem[Venn et al.(2004)]{venn04} Venn, K.~A., Irwin, M., 

1382: Shetrone, M.~D., Tout, C.~A., Hill, V., \& Tolstoy, E.\ 2004, \aj, 128, 

1383: 1177  

1384: 

1385: \bibitem[Vivas \& Zinn(2006)]{vivas06} Vivas, A.~K., \& Zinn, 

1386: R.\ 2006, \aj, 132, 714 

1387: 

1388: \bibitem[Vivas et al.(2005)]{vivas05} Vivas, A.~K., Zinn, R., 

1389: \& Gallart, C.\ 2005, \aj, 129, 189 

1390: 

1391: \bibitem[Willman et al.(2002)]{willman02} Willman, B., Dalcanton, 

1392: J., Ivezi{\'c}, {\v Z}., Jackson, T., Lupton, R., Brinkmann, J., Hennessy, 

1393: G., \& Hindsley, R.\ 2002, \aj, 123, 848 

1394: 

1395: \bibitem[White \& Springel(2000)]{ws00} White, S.~D.~M.~\& Springel,

1396: V.\ 2000, in The First Stars, ed. A. Weiss, T.~G. Abel, V. Hill (Berlin:

1397: Springer--Verlag), 327

1398: 

1399: %\bibitem[Yoss, Neese, \& Hartkopf(1987)]{ynh87}Yoss, K.~M., Neese, C.~L., \&

1400: %Hartkopf, W.~I. 1987, \aj, 94, 1600

1401: 

1402: 

1403: \bibitem[Zinn(1985)]{z85}Zinn, R. 1985, \apj, 293, 424

1404: 

1405: \end{thebibliography}

1406: 

1407: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1408: 

1409: %%%%%%%%%%%%%%%

1410: %   TABLES    %

1411: %%%%%%%%%%%%%%%

1412: 

1413: %% Table 1

1414: %%\input{tables/meta_super_duper_newtrans_newrr_err.tex}

1415: \input{tab1.tex}

1416: 

1417: \clearpage

1418: 

1419: %% Table 2

1420: \begin{deluxetable}{crr}

1421: \tablewidth{0pt}

1422: \tablecaption{Smooth Halo Model Parameters \label{smo_params}}

1423: \tablehead{ 

1424:   \colhead{} &

1425:   \colhead{With H99 Streams} &

1426:   \colhead{Without H99 Streams} }

1427: \tablecolumns{3}

1428: \startdata

1429: $\avevR$	&  $ -0.93 \pm 10.20$	& $ -3.57 \pm 10.63$ \\

1430: $\sigma_R$	&  $154.96 \pm 7.21$	& $157.30 \pm 7.52$ \\

1431: $\avevphi$	&  $ 28.62 \pm 7.19$	& $23.42 \pm 7.41$ \\

1432: $\sigma_\phi$	&  $109.33 \pm 5.09$	& $109.64 \pm 5.24$ \\

1433: $\avevz$	&  $ -6.43  \pm 6.66$	&   $-1.32 \pm 5.66$ \\     

1434: %% BELOW is Z FLIP (Z pointing in direction of SGP).

1435: %%$\avevz$	&  $ 6.43  \pm 6.66$	&   $1.32 \pm 5.66$ \\     

1436: $\sigma_z$	&  $101.23  \pm 4.71$	&  $83.75 \pm 4.00$ \\

1437: \enddata

1438: \end{deluxetable}

1439: 

1440: 

1441: %% \begin{deluxetable}{ccccccc}

1442: %% \tablewidth{0pt}

1443: %% \tablecaption{Smooth Halo Model Parameters \label{smo_params} }

1444: %% \tablehead{ \colhead{Data Set} & \colhead{$\avevR$} & \colhead{$ 

1445: %% \sigma_R$} & \colhead{$\avevphi$} & \colhead{$\sigma_\phi$} & \colhead 

1446: %% {$\avevz$} & \colhead{$\sigma_z$ } }

1447: %% \startdata

1448: %% Combined Data  with Stream & $1.59 \pm 9.94$  &     $152.11 \pm 7.03$  &  $30.53 \pm 7.60$   &    $116.22 \pm 5.37$ &  $6.86 \pm 6.62$     &   $101.20 \pm 4.68$  \\

1449: %% Combined Data without Stream &  $5.84 \pm 10.63$   &    $151.40 \pm  7.51$  & $19.58 \pm 6.64$    &  $94.56 \pm 4.69$ &  $-1.03 \pm 5.52$   &    $78.65 \pm 3.90$     \\

1450: %% \enddata

1451: %% \end{deluxetable}

1452: 

1453: %% Table 3

1454: \begin{deluxetable}{lll}

1455: \tablewidth{0pt}

1456: \tablecaption{Shapiro-Wilk Results for H99 data \label{sw_h99} }

1457: \tablehead{ \colhead{Data Set} & $N_{stars}$ & \colhead{p-value} }

1458: \startdata

1459: All Stars			&  101 &  0.009 \\

1460: Excluding H99 Streams		&  94  & 0.05 \\

1461: Excluding H99 Streams \& HD 124358	& 93   & 0.19    \\

1462: \enddata

1463: \end{deluxetable}

1464: 

1465: \clearpage

1466: 

1467: %% Table 4

1468: %%\input{tables/meta_strm_table.tex}

1469: \input{tab4.tex}

1470: 

1471: %% Table 5

1472: %%\input{tables/meta_out1_table.tex}

1473: \input{tab5.tex}

1474: 

1475: %% Table 6

1476: %%\input{tables/meta_out2_table.tex}

1477: \input{tab6.tex}

1478: 

1479: \clearpage

1480: 

1481: %% Table 7

1482: %%\input{tables/box_prob_table.tex}

1483: \input{tab7.tex}

1484: 

1485: %% Table 8

1486: %%\input{tables/meta_newrr_newtrans_vz_results.tex}

1487: \input{tab8.tex}

1488: 

1489: %% Table 9

1490: %%\input{tables/meta_newrr_newtrans_vphi_results.tex}

1491: \input{tab9.tex}

1492: 

1493: \clearpage

1494: 

1495: 

1496: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1497: 

1498: %%%%%%%%%%%%%%%

1499: %   FIGURES   %

1500: %%%%%%%%%%%%%%%

1501: 

1502: %% Figure 1

1503: \begin{figure}

1504: \centering

1505: %%\includegraphics[scale=0.65,angle=-90]{figures/boo_cy_dist.v2.eps}

1506: \includegraphics[scale=0.65,angle=-90]{f1.eps}

1507: \caption{Difference between distance estimates for red giants from ATT94

1508: and B00 as a function of distance. The differences are expressed as a

1509: fraction of the ATT94 distance. It can be seen that for stars closer

1510: than 1 kpc, the B00 distances are on average 40\% smaller than the ATT94

1511: distances.}

1512: \label{boo_cy_dist}

1513: \end{figure}

1514: 

1515: %% Figure 2

1516: \begin{figure}

1517: \centering

1518: %%\includegraphics[scale=0.70,angle=-90]{figures/boo_cy_bv.v2.eps}

1519: \includegraphics[scale=0.70,angle=-90]{f2.eps}

1520: \caption{Difference between red giant distance estimates from ATT94 and

1521: B00, shown as a function of (B--V)$_0$ color. It can now be seen that

1522: the problematic stars are the bluest ones, with (B--V)$_0 < 0.9$.}

1523: \label{boo_cy_bv}

1524: \end{figure}

1525: 

1526: %% Figure 3

1527: \begin{figure}

1528: \centering

1529: \includegraphics[scale=0.7,angle=-90]{f3.eps}

1530: \caption{Distribution of distance and absolute value of Galactic

1531: latitude ($b$) for (a) the RR Lyrae variables in our sample and (b) the

1532: red giant and RHB stars in our sample.}

1533: \label{lb}

1534: \end{figure}

1535: 

1536: 

1537: %% Figure 4

1538: \begin{figure}

1539: \centering

1540: %%\includegraphics[scale=0.7]{figures/newsim_ang_mom_amina.ps}

1541: \includegraphics[scale=0.7]{f4.eps}

1542: \caption{Structure in velocity and angular momentum space for the H99

1543:   star streams model for stars within 2.5~kpc of the Sun. The bottom

1544:   panel is a plot of the angular momentum in the plane of the Galaxy's

1545:   disk as a function of angular momentum out of the disk. The top

1546:   panels show the cylindrical velocity coordinates of the data plotted

1547:   against each other.}

1548: \label{ang_mom_am}

1549: \end{figure}

1550: 

1551: %% Figure 5

1552: \begin{figure}

1553: \centering

1554: %%\includegraphics[scale=0.7]{figures/ang_mom_am_smo_meta.ps}

1555: \includegraphics[scale=0.7]{f5.eps}

1556: \caption{Structure in velocity and angular momentum space for the

1557: smooth model for stars within 2.5~kpc of the Sun.}

1558: \label{ang_mom_smooth}

1559: \end{figure}

1560: 

1561: %% Figure 6

1562: \begin{figure}

1563: \centering

1564: %%\includegraphics[scale=0.7]{figures/ang_mom_h99_v2.ps}

1565: \includegraphics[scale=0.7]{f6.eps}

1566: \caption{Plots of the distribution of the H99 data in velocity and

1567: angular momentum space for stars with $\feh \leq -1.6$ and distances

1568: less than 1.0 kpc from the Sun.  The

1569: stream stars are represented by the large circles. The additional

1570: $v_z$ outlier (HD 124358) is shown by the upward pointing triangle. }

1571: \label{ang_mom_h99}

1572: \end{figure}

1573: 

1574: %% Figure 7

1575: \begin{figure}

1576: \centering

1577: %%\includegraphics[scale=0.7]{figures/vz_hist_h99_v2.ps}

1578: \includegraphics[scale=0.7]{f7.eps}

1579: \caption{Histogram of the H99 data $v_z$ velocity distribution with

1580:   the stream stars shaded. A normal distribution is plotted as a

1581:   dashed line. Note how the H99 stream stars widen the wings of the

1582:   velocity distribution making it deviate from normality.}

1583: \label{vz_hist_h99}

1584: \end{figure}

1585: 

1586: %% Figure 8

1587: \begin{figure}

1588: \centering

1589: %%\includegraphics[scale=0.7]{figures/frac_stream_plot.ps}

1590: \includegraphics[scale=0.7]{f8.eps}

1591: \caption{Fraction of stream stars in halo as a function of distance

1592: limit. Note that the fraction of stream stars decreases as the

1593: distance limit increases when the fraction of stream stars is

1594: normalized to 10\% of the halo stars within 1~kpc of the Sun.}

1595: \label{frac_stream_plot}

1596: \end{figure}

1597: 

1598: %% Figure 9

1599: \begin{figure}

1600: \centering

1601: %%\includegraphics[scale=0.7]{figures/n_pw_dist_plot_2.ps}

1602: \includegraphics[scale=0.7]{f9.eps}

1603: \caption{Average p-value for 10,000 random samples of $v_z$ velocities

1604: with the relative number of stream to smooth halo stars fixed at 10\%

1605: for a distance limit of 1.0 kpc. The standard deviation of the trial

1606: p-values is on the order of the width of the lines. The distance

1607: limits used for the samples are 1.0, 2.0, 3.0, 4.0, and 5.0 kpc. As

1608: the distance limit increases, the size of the sample for which

1609: detections are possible also increases.}

1610: \label{n_pw_dist_plot}

1611: \end{figure}

1612: 

1613: %% Figure 10

1614: \begin{figure}

1615: \centering

1616: %%\includegraphics[scale=0.7]{figures/n_pave_dist_b_newsim.ps}

1617: \includegraphics[scale=0.7]{f10.eps}

1618: \caption{Average p-value for 10,000 random samples of radial

1619: velocities (corrected to the Galactocentric frame of reference) with

1620: 10\% stream stars for a distance limit of 1.0~kpc.  The standard

1621: deviation of the trial p-values is on the order of the width of the

1622: lines. Galactic latitude limits ($|b| \ge 40$, $|b| \ge 50$, $|b| \ge

1623: 60$, $|b| \ge 70$, and $|b| \ge 80$) were imposed on samples with five

1624: different distance limits (1.0, 2.0, 3.0, 4.0, and 5.0 kpc). N.B. The

1625: x-axes on these graphs all have different scales, since the range of

1626: samples tested is very different in each case.}

1627: \label{n_pave_dist_b}

1628: \end{figure}

1629: 

1630: %% Figure 11

1631: \begin{figure}

1632: \centering

1633: %%\includegraphics[scale=1.0]{figures/meta_newtrans_fehm10_d25_ang_mom_groups_color.ps}

1634: \includegraphics[scale=1.0]{f11.eps}

1635: \caption{Angular momentum and cylindrical velocity plots for the

1636: combined data sample with the selection criteria: $ D \leq 2.5 \

1637: \mathrm{kpc}$ and $\feh \leq -1.0$. The retrograde outliers are

1638: indicated by downward0facing triangles, the prograde outliers by

1639: upward-facing triangles, and the H99 streams by large circles. Note

1640: that the empty region of the angular momentum diagram at $J_z =

1641: 1500-2500 \ \rm{kpc} \ \kms, J_\perp < 600 \ \rm{kpc} \ \kms$ was

1642: occupied by likely thick-disk stars.  These stars were excluded from

1643: our final sample based on their position in the angular momentum

1644: diagram.}

1645: \label{meta_fehm10_d25_ang_mom_groups}

1646: \end{figure}

1647: 

1648: %% Figure 12

1649: \begin{figure}

1650: \centering

1651: %%\includegraphics[scale=1.0]{figures/model_vz_evolution.eps}

1652: \includegraphics[scale=1.0]{f12.eps}

1653: \caption{Probability that the stream with negative $v_z$ will make

1654:    up a given fraction of the total number of stars from the accreted

1655:    satellite near the Sun calculated using the H99 model at four

1656:    different times: 3, 6, 9, and 12 Gyr. The dashed line

1657:    indicates the observed fraction of stars in this stream ($8/11 =

1658:    0.72$).}

1659: \label{model_vz_evolution}

1660: \end{figure}

1661: 

1662: %% Figure 13

1663: \begin{figure}

1664: \centering

1665: %%\includegraphics[scale=0.7]{figures/gal_coords_mb61.eps}

1666: \includegraphics[scale=0.7]{f13.eps}

1667: \caption{H99 model $(x,y,z)$ and the local, Sun-centered

1668:   $(x',y',z')$ coordinate systems. Note that both the H99 model

1669:   coordinate system and the local, Sun-centered coordinate system are

1670:   left-handed. The direction of Galactic rotation is in the positive

1671:   $y'$-direction.}

1672: \label{gal_coords}

1673: \end{figure}

1674: 

1675: \end{document}

1676: