1: %\documentclass[12pt,onecolumn,manuscript]{aastex}

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

3: %\documentclass[preprint2]{aastex}

4: %\usepackage{emulateapj5}

5: 

6: %\doublespace

7: %\onecolumn

8: 

9: 

10: \newcommand{\etal}{{et~al.~}}

11: \newcommand{\fcrao}{\mbox{$\rm FCRAO$}}

12: \newcommand{\IRAS}{\mbox{$\rm IRAS$}}

13: \newcommand{\HIRES}{\mbox{$\rm HiRes$}}

14: \newcommand{\tb}{\mbox{$\rm T_B$~}}

15: \newcommand{\co}{\mbox{$^{12}$CO}}

16: \newcommand{\coa}{\mbox{$^{13}$CO}}

17: \newcommand{\wco}{\mbox{W$_{CO}$~}}

18: \newcommand{\lco}{\mbox{L$_{CO}$}}

19: \newcommand{\cc}{\mbox{${\rm cm}^{-3}$}}

20: \newcommand{\cmsq}{\mbox{${\rm cm}^{-2}$}}

21: \newcommand{\kms}{\mbox{${\rm km~s}^{-1}$}}

22: \newcommand{\htwo}{\mbox{${\rm H}_2$}}

23: \newcommand{\msun}{\mbox{$M_\odot$~}}

24: \newcommand{\lsun}{\mbox{$L_\odot$~}}

25: \newcommand{\hr}{\mbox{$\rm^{h}$}}

26: \newcommand{\mm}{\mbox{$\rm^{m}$}}

27: \newcommand{\sigmasqr}{\mbox{${\sigma}^2$}}

28: %%\newcommand{\arcsec}{\mbox{$\rm^{'}$}}

29: \received{}

30: %\accepted{}

31: %\journalid{}{}

32: %\articleid{}{}

33: 

34: %\slugcomment{DRAFT \today}

35: 

36: % Authors may supply running head information, if they wish to do so, although

37: % this may be modified by the editorial offices.  The left head contains a

38: % list of authors, usually three allowed---otherwise use et al.  The right

39: % head is a modified title of up to roughly 44 characters.  Running heads

40: % are not printed.

41: 

42: \lefthead{Heyer \& Brunt}

43: \righthead{The Universality of Turbulence in Galactic Molecular Clouds} 

44: 

45: % This is the end of the "preamble".  Now we wish to start with the

46: % real material for the paper, which we indicate with \begin{document}.

47: % Following the \begin{document} command is the front matter for the

48: % paper, viz., the title, author and address data, the abstract, and

49: % any keywords or subject headings that are relevant.

50: 

51: \begin{document}

52: 

53: \def\gtabouteq{\,\hbox{\raise 0.5 ex \hbox{$>$}\kern-.77em

54:                     \lower 0.5 ex \hbox{$\sim$}$\,$}}

55: \def\ltabouteq{\,\hbox{\raise 0.5 ex \hbox{$<$}\kern-.77em

56:                      \lower 0.5 ex \hbox{$\sim$}$\,$}}

57: \def\vlsr{V$_{LSR}$}

58: \def\kms{km s$^{-1}$}

59: %%\def\etal{{et al.}}

60: 

61: %\title{The Common Dynamical State of Galactic Molecular Clouds} 

62: \title{The Universality of Turbulence in Galactic Molecular Clouds} 

63: 

64: \author{Mark H. Heyer\altaffilmark{1}, 

65: Christopher M. Brunt \altaffilmark{1}

66: }

67: 

68: \altaffiltext{1}{Department of 

69: Astronomy, Lederle Research Building,

70: University of Massachusetts, Amherst, MA 01003, USA;

71: heyer@astro.umass.edu,brunt@astro.umass.edu}

72: 

73: \begin{abstract}

74: The universality of interstellar 

75: turbulence is examined from observed structure functions of 27 giant 

76: molecular clouds and Monte Carlo modeling.  

77: We show that the structure functions, ${\delta}v={v_\circ}l^\gamma$, 

78: derived from 

79: wide field imaging of \co\ J=1-0 emission from 

80: individual clouds are described by a narrow range in the scaling exponent, 

81: $\gamma$, and the scaling coefficient, $v_\circ$. 

82: The similarity of turbulent structure functions emphasizes the universaility 

83: of turbulence in the molecular interstellar medium and accounts for 

84: the cloud-to-cloud size-line width relationship

85: initially identified by Larson (1981).  

86: The degree of turbulence universality is 

87: quantified by Monte Carlo simulations that reproduce the mean 

88: squared velocity residuals 

89: of the observed cloud-to-cloud relationship.  

90: Upper limits to the variation of the 

91: scaling amplitudes and exponents for molecular clouds are $\approx$10-20\%.  

92: The 

93: measured invariance of turbulence for molecular clouds with vastly 

94: different sizes, environments, and star formation activity suggests 

95: a common formation mechanism such as converging turbulent flows 

96: within the diffuse ISM and a limited contribution of energy from 

97: sources within the cloud 

98: with respect to large scale driving mechanisms.

99: \end{abstract}

100: \parindent=20pt

101: \keywords{ISM: clouds - ISM: kinematics and dynamics -ISM: molecules - ISM: structure} 

102: 

103: \section{Introduction}

104: 

105: Turbulent, non-laminar gas flows are a ubiquitous feature within 

106: all phases of the interstellar medium (ISM).  Therefore,

107: accurate descriptions of interstellar turbulence are essential to 

108: meaningful understanding of ISM dynamics and star formation.  

109: An important statistical description of fluid dynamics is the generalized 

110: velocity structure function, 

111: $$ S_p(l)=<|v(r)-v(r+l)|^p> \eqno(1), $$ 

112: where $l$ is the spatial displacement between two cells within a 3 dimensional 

113: volume, $p$ is the order of the structure function,

114: and the averages are taken over the volume of the fluid.  Over a finite 

115: spatial range, the structure functions may be described as a power law, 

116: $S_p(l) \sim l^{\zeta(p)}$ but is often re-framed as an equivalent linear 

117: expression by taking the pth root, $S_p(l)^{1/p}={\delta}v={v_\circ}l^\gamma$,

118: where $\gamma=\zeta(p)/p$ is the scaling exponent and $v_\circ$ is the scaling 

119: coefficient.

120: The structure function provides a concise description of the 

121: spatial coherence of velocity differences

122: within a given region.  Such differences

123: can arise from both systematic motions (rotation, collapse, outflows) and 

124: random motions due to turbulent gas flows.  Within interstellar clouds, most 

125: velocity differences are due to turbulence.

126: 

127: One of the most 

128: cited and influential studies on interstellar turbulence is Larson (1981) that 

129: identified a power law relationship between 

130: the global velocity dispersion, ${\Delta}v$ (\kms), and cloud size, $L$ (pc), 

131: of molecular clouds 

132: from values taken from 

133: the literature, ${\Delta}v=CL^{\Gamma}$ where $C=1.1$ and

134: $\Gamma=0.38$.  

135: %%Different variables are used to formally distinguish the 

136: %%structure function of an individual cloud from the cloud-to-cloud size line

137: %%width relationship.

138: %Larson (1981) interpreted 

139: %this relationship as a signature of 

140: %Kolmogorov 

141: %turbulence for which $\gamma$=1/3. 

142: Using a more homogeneous set of cloud data from the 

143: Massachusetts-Stony Brook Galactic Plane Survey, Solomon \etal (1987) 

144: found a similar correlation with a comparable scaling coefficient 

145: ($C=1.0$) but a steeper scaling exponent ($\Gamma=0.5$).

146: These relationships rely on global velocity dispersions that are distinguished

147: from velocity differences codified in the structure function of equation (1).

148: As the sample clouds are 

149: distributed throughout the Galaxy, 

150: these do not comprise a singular fluid volume.

151: Therefore, the connection of these cloud-to-cloud size-line width relationships to 

152: velocity structure functions of individual clouds would seem remote  

153: unless one assumes that all clouds 

154: have approximately the same values for $v_\circ$ and $\gamma$. 

155: Larson (1981) showed examples of a similar 

156: scaling law {\it within clouds} by using sizes and velocity dispersions derived from 

157: molecular tracers with different excitation requirements 

158: (see also Fuller \& Myers 1992).  However, 

159: the 

160: dense regions comprise a small fraction of the mass and 

161: volume of a molecular 

162: cloud so the measured velocity dispersions within the spatial scales of 

163: CS or NH$_3$ emissions may not be 

164: representative of velocity differences over comparable scales but within a lower

165: density substrate.  Size-line width relationships derived from clump 

166: identification algorithms for individual clouds 

167: exhibit a large degree of scatter with a range of values for $\gamma$ 

168: (Carr 1987; St\"utzki 

169: \& G\"usten 1990; Falgarone, Puget, 

170: \& Pereault 1992) or no 

171: relationship at all (Loren 1989; Simon etal 2001).  

172: The absence of a consistent correlation between size and line width of 

173: embedded structures within a given cloud is due to the limited dynamic 

174: range of sizes that can be identified by such algorithms 

175: and the superposition of emission 

176: from disconnected features along the line of sight (Ballesteros-Paredes \&

177: Mac Low 2002).

178: Using structure functions of velocity centroid images, 

179: Meisch \& Bally (1994)

180: showed that the scaling exponents are similar for a sample of 4 

181: molecular clouds but did not consider the variation of the scaling coefficient.

182: Brunt (2003) 

183: provides the most convincing evidence for 

184: similar scaling laws within molecular clouds using Principal Component 

185: Analysis (PCA) as a tool to recover the true structure function for a given cloud 

186: (Brunt \& Heyer 

187: 2002).  For each \co\ or \coa\  spectroscopic data cube of a molecular cloud, a set of 

188: ${\delta}v$,$l$ points are determined from the eigenvectors 

189: and eigenimages.  When the PCA measurements for all clouds are 

190: combined onto a single plot,

191: these define

192: a nearly co-linear set of points.  Such a correlation necessarily results from 

193: narrow distributions of the 

194: scaling exponent, $\gamma$,  and coefficient, $v_\circ$, 

195: for this sample of clouds. 

196: 

197: In this {\it Letter}, we extend the study of Brunt (2003) to 

198: demonstrate that {\it Larson's cloud-to-cloud 

199: scaling law is explained only if the structure functions for individual 

200: clouds are nearly identical.} 

201: PCA-based ${\delta}v,l$ relationships 

202: are presented 

203: that demonstrate the same functional form for structure functions 

204: for molecular clouds that span 

205: a wide range in size and environmental conditions.  Monte Carlo models

206: are constructed that place upper limits to the variation of the 

207: scaling coefficient and exponent.  Finally, we discuss the consequences

208: of an invariant turbulent spectrum in the context of the formation 

209: of interstellar molecular clouds, sources of turbulent energy, and star 

210: formation.

211: 

212: \section{The Composite Structure Function}

213: 

214: Following Brunt \& Heyer (2002), PCA is applied to spectroscopic data cubes

215: of \co\ J=1-0 emission from molecular clouds that are part of recent 

216: wide field imaging surveys at the Five College Radio 

217: Astronomy Observatory (Heyer \etal 1998; Brunt \& Mac Low 2004) or targeted 

218: studies of individual giant molecular clouds.  

219: %Despite the large optical depths, 

220: %\co\ emission provides the best tracer of the large scale dynamics of a molecular

221: %cloud.  

222: Heyer \& Schloerb (1997) and Brunt (2003) show there is little difference 

223: in the ${\delta}v,l$ relationships derived from \co\ emission and 

224: the lower opacity \coa\ emission.

225: For each cloud, a power law is fit to the ${\delta}v,l$ points to determine 

226: the PCA scaling exponent, $\alpha_{PCA}$,  and coefficient, $v_\circ$.  

227: For the sample of 27 molecular

228: clouds, the mean and standard deviation for 

229: the scaling exponent are 0.62 and 0.09 respectively.  

230: Based on models with little or zero intermittency, 

231: this  

232: PCA scaling exponent corresponds to a 

233: structure function exponent equal to $0.49\pm0.15$ (Brunt \etal 2003).  

234: The 

235: mean and standard deviation of the scaling 

236: coefficient are 0.90 \kms\  and 0.19 \kms.

237: These rather narrow distributions of $\gamma$ and $v_\circ$ re-emphasize 

238: the results of Brunt (2003) that there is not much variation in the 

239: structure function parameters between molecular clouds.

240: In Figure~1, we overlay the PCA ${\delta}v,l$ points from the sample of clouds.

241: The composite points 

242: reveal a near-identical form of the inferred structure functions.  The solid line 

243: shows the power law bisector fit to 

244: all points, ${\delta}v=(0.87\pm0.02)l^{0.65\pm0.01}$.  This PCA derived 

245: exponent corresponds to a structure function scaling exponent of 0.56$\pm$0.02.

246: 

247: The global velocity dispersion of 

248: each cloud and the cloud size are determined from the scales of the 

249: first eigenvector and eigenimage respectively.

250: Basically, the global velocity dispersion, ${\Delta}v$, is the value of the 

251: velocity structure function measured at the size scale, $L$, of the cloud.

252: These points, marked as filled circles within Figure~1, 

253: are equivalent to the global values used by 

254: Larson (1981) and Solomon \etal (1987) that define the 

255: cloud-to-cloud size-line width relationship.  

256: A power law bisector to this subset  

257: of points is ${\Delta}v=(0.96\pm0.17)L^{0.59\pm0.07}$. 

258: The similarity of this

259: cloud-to-cloud 

260: relationship with that of the composite points is a

261: consequence of the uniformity of the 

262: individual structure functions.  Within the quoted errors, it is 

263: also similar to the cloud-to-cloud 

264: size-line width relationships -- $\gamma \sim \Gamma$ and $v_\circ \sim C$.  

265: Therefore, {\it Larson's 

266: global velocity dispersion versus cloud 

267: size scaling law follows directly 

268: from the near identical functional form of velocity structure functions

269: for all clouds.}

270: If there were 

271: significant differences of $\gamma$ or $v_\circ$ between clouds, 

272: then the cloud-to-cloud size-line width relationship would exhibit much larger

273: scatter than is measured by Larson (1981) and Solomon \etal (1987).

274: 

275: \section{The Degree of Turbulence Universality}

276: 

277: The cloud-to-cloud size-line width relationships measured by Larson (1981)

278: and Solomon \etal (1987) and the composite structure functions shown in 

279: Figure~1 do exhibit some degree of scatter about the fitted lines.  

280: The scatter is quantified by the mean square of the velocity residuals, ${\sigma}_{obs}^2$,

281: for each data set

282: where

283: $$ {\sigma}_{obs}^2 = { {\Sigma_i^{N} ({\Delta}v_i-CL_i^\Gamma)^2} \over {N} } \;\;\;\; km^2s^{-2} \eqno(2), $$

284: N is the number of clouds in the sample, C  and $\Gamma$ are 

285: the parameters derived by fitting a power law to the observed ${\Delta}v,L$ 

286: points. 

287: The value for ${\sigma}_{obs}^2$ for the sample of clouds in Larson (1981) using 

288: only the \co\ and \coa\ measurements is 1.41 $km^2 s^{-2}$.  

289: The Solomon \etal (1987) 

290: sample is a larger, more homogeneous set of clouds and therefore, provides a

291: more accurate measure of the variance within the cloud-to-cloud size-line 

292: width relationship.  The corresponding

293: ${\sigma}_{obs}^2$ is 0.88 $km^2 s^{-2}$.   The value of 

294: ${\sigma}_{obs}^2$ for the ${\Delta}v,L$ points in Figure~1 is 1.93 $km^2 s^{-2}$

295: and 0.35 $km^2 s^{-2}$ for the composite collection of ${\delta}v,l$ points.

296: 

297: The measured scatter, described by ${\sigma}_{obs}^2$, 

298: of the size-line width relationships is a 

299: critical constraint to the degree of invariance of turbulence within 

300: the molecular interstellar medium. 

301: The scatter arises from several sources.  There are basic measurement errors

302: in the global velocity dispersion owing to the 

303: velocity resolution of the measurements and the cumulative statistical error 

304: of the individual spectra.  Deriving cloud sizes from complex projected 

305: distributions 

306: of the molecular gas may also introduce some scatter.  These measurement

307: errors are rarely shown in any cloud size-line width plots.

308: A secondary source of scatter is limited or biased

309: mapping of the molecular cloud.  If a given  map was limited in angular extent 

310: and centered on a region within the cloud that is actively forming stars

311: %\footnote{Such biased mapping was typical for 

312: %many of the early maps of molecular clouds}

313: then 

314: the measured "global" velocity dispersion 

315: may be broader due to localized, expanding motions 

316: from HII regions or protostellar outflows.  Such enhanced velocity 

317: dispersions from biased regions may not 

318: represent the global velocity dispersion of the entire giant molecular 

319: cloud, that is typically quite extended with respect to star formation sites

320: within the cloud.  

321: An additional source of the observed scatter is the uncertainty of the 

322: distance to each cloud in the sample.   Distances to molecular clouds,

323: and correspondingly, cloud sizes, 

324: are generally not known to

325: precisions better than 25\%.

326: Finally, and most important for the subject of this study,

327: true differences in the scaling coefficient and exponent

328: would also contribute to 

329: the observed scatter in the size-line width relationship.  It is this 

330: component that we wish to constrain as it defines the degree to which 

331: turbulence is invariant in the ISM.

332: 

333: To gauge the universality of turbulence within the molecular interstellar 

334: medium, a simple, Monte Carlo model is constructed to place limits on the 

335: variance of the scaling coefficients and scaling exponents of 

336: individual clouds.  The structure function for each model cloud is defined by 

337: the parameters, $v_\circ=<v_\circ>+\epsilon_v$ and 

338: $\gamma=<\gamma>+\epsilon_\gamma$

339: where $\epsilon_{v}$ and $\epsilon_{\gamma}$ are drawn from gaussian 

340: probability distributions with 

341: standard deviations of $\sigma_v$ and $\sigma_\gamma$ respectively.

342: Both $\sigma_v$ and $\sigma_\gamma$ implicitly contain contributions from 

343: the measurement errors, biased mapping, and true variations in turbulence.

344: In the limit of absolute universal turbulence, infinite precision, 

345: and 

346: unbiased imaging, $\sigma_v$ = $\sigma_\gamma$=0.

347: A size, $L_\circ$, is assigned to each cloud from a 

348: uniform probability distribution

349: such that 

350: $0 < log(L_{\circ}) < 2$, corresponding to cloud sizes between 1 and 100 pc.

351: A global velocity dispersion, ${\Delta}v$, is determined by evaluating 

352: the structure function at the assigned size of the cloud,

353: $$ {\Delta}v = (<v_\circ>+\epsilon_v)L_\circ^{<\gamma>+\epsilon_{\gamma}} \eqno(3) $$

354: Once ${\Delta}v$ is determined from equation (3),  

355: a random component, $\epsilon_{L}$, 

356: is added to the cloud size, $L=L_{\circ}(1+\epsilon_{L})$ 

357: to replicate an uncertainty in cloud distances where 

358: $\epsilon_{L}$ is drawn from a uniform probability distribution within the 

359: percentage range, ${\pm}\sigma_D/D$.

360: Simulations are run for ${\sigma}_D/D$=0, 0.1, 0.25, and 0.5.

361: The Solomon \etal (1987) data set

362: is used as the primary 

363: observational constraint so 

364: N=272.  Following the results in Figure~1, 

365: %\footnote{The original sample of Solomon contains 273 clouds 

366: %but one entry in their

367: %table contains a size with 0.00 degree so we exclude 

368: %this cloud from our 

369: %analysis.} 

370: $<v_\circ>$=0.9 \kms, and $<\gamma>$=0.5.  For each set of ${\sigma}_v$ 

371: and $\sigma_\gamma$ parameters, \sigmasqr\ is calculated,

372: $${\sigma}^2(\sigma_\gamma,\sigma_v) = { {\Sigma_i^N 

373: ({\Delta}v_i - C_1L_i^{\Gamma_1})^2} \over {N} } \;\;\;\; km^2s^{-2} \eqno(4), $$

374: where $C_1$ and $\Gamma_1$ are determined from a bisector fit to

375: the ${\Delta}v,L$ points for a single realization. 

376: To reduce the statistical errors of the simulation, we calculate the mean value 

377: of ${\sigma}^2(\sigma_\gamma,\sigma_v)$ from 500 realizations.

378: 

379: The Monte Carlo results are shown in Figure~2 where 

380: $\sigma_v$ and $\sigma_\gamma$ are normalized by  $<v_\circ>$ and $<\gamma>$ 

381: respectively to 

382: reflect the fractional variation.

383: The heavy solid line shows the value of ${\sigma}_{obs}^2$=0.88 $km^2s^{-2}$

384: from the Solomon \etal (1987)

385: sample.

386: This contour 

387: defines the locus of $\sigma_v/<v_\circ>$, $\sigma_\gamma/<\gamma>$ 

388: points that reproduce the observed scatter in the Solomon \etal (1987) 

389: size-line width relationship.  In the unlikely extreme 

390: limit that $\sigma_\gamma$=0,

391: then the most $\sigma_v$ can vary is 18-23\% about the observed value of 0.9

392: \kms.  Conversely, if $\sigma_v$=0, then $\sigma_\gamma$ can vary 

393: by less than 9-12\% about its fiducial value of 0.5.  

394: More realistically, $\sigma_v \neq 0$ and $\sigma_\gamma \neq 0$, 

395: so the percent variations for both parameters are $\sim$8-12\%. 

396: We emphasize that these are {\it upper limits} to the true 

397: variations between clouds

398:  as 

399: $\sigma_v$ and  $\sigma_\gamma$ 

400: also include measurement errors and 

401: biased mapping effects that are likely present in all cloud-to-cloud size-line 

402: width relationships.

403: 

404: \section{Implications of Invariant Interstellar Turbulence}

405: 

406: The upper limits to variations of the structure function parameters 

407: are quite small when one considers the large range 

408: of molecular cloud environments.  With few exceptions,

409: giant molecular 

410: clouds are sites of massive star formation.  The massive stars can, 

411: in principle, affect

412: gas dynamics over the extent of a cloud by enhancing the UV radiation field,

413: driving HII region 

414: expansions and stellar winds, and are the progenitors of 

415: supernova explosions.  The star formation activity in smaller clouds is 

416: generally limited to the birth of low mass stars 

417: whose impact on cloud dynamics by protostellar winds 

418: is highly localized and small with respect to the integrated energy input 

419: from massive stars.

420: For the sample of molecular clouds used in this study, the 

421: far infrared {\it IRAS} point source luminosities range from 20 \lsun\ 

422: (B18, Heiles' Cloud 2, L1228) to 

423: 7.6$\times$10$^5$ \lsun\ (NGC 7538).

424: Despite the large differences in the amount 

425: of internal energy injection from newborn stars, 

426: the functional forms of the turbulent 

427: structure functions for all clouds are similar.

428: Either all clouds coincidentally 

429: redistribute this internal energy into turbulent, 

430: random motions described by structure functions with the same slopes and 

431: amplitudes via some self-regulatory process 

432: or these 

433: internal energy sources are small compared to an external energy reservoir 

434: that is common to all regions in the Galaxy.  

435: Brunt \etal (2004) show that most of the 

436: turbulent energy of a molecular cloud resides on the largest scales

437: and that the cloud dynamics are more readily accounted by large 

438: scale driving of turbulence.  The universality of turbulence described 

439: in this study provides additional evidence for large scale driving 

440: sources within the molecular interstellar medium.  The narrow range 

441: of turbulent flow parameters may also reflect the necessary conditions that 

442: facilitate the development of molecular clouds.  Regions with extreme,

443: high values

444: of $v_\circ$ may be highly overpressured with respect to self-gravity and 

445: expand to larger, more diffuse configurations with less effective 

446: self-shielding to sustain significant molecular abundances.

447: 

448: Larson (1981) presaged the results from recent numerical simulations 

449: that molecular clouds are 

450: short lived ($<$10$^7$ years), transient objects (Ballesteros-Paredes, 

451: Hartmann, \& Vazquez-Semadeni 1999) that must 

452: be continually reassembled from the diffuse gas component.  His suggestion 

453: that molecular clouds result from thermal instabilities within shocks 

454: of colliding atomic gas streams is 

455: similar to recent numerical simulations of compressible turbulence that show

456: the development of high density regions directly from the shocks

457: (Hunter \etal 1986; Elmegreen 1993; Ballesteros-Paredes, Vazquez-Semadeni, \& Scalo 1999).

458: Such a common dynamical origin of 

459: molecular clouds in the Galaxy may account for the measured 

460: near invariance of turbulence.

461: 

462: The results presented in this study apply to the low density gas substrate

463: that comprises most of the mass and volume of a molecular cloud.  Turbulent 

464: properties may indeed differ between clouds 

465: within the high density, localized, supercritical 

466: core 

467: regions.   In distributed,  low mass star forming regions, the non-thermal 

468: motions within the dense gas are 

469: subsonic (Benson \& Myers 1989).  Within the massive cores typical 

470: of clustered star forming 

471: regions, the observed velocities 

472: are 

473: supersonic (Pirogov \etal 2003).

474: The identification of processes responsible for such divergent, 

475: dense gas configurations 

476: remains one of the primary 

477: challenges to descriptions of star formation 

478: (Klessen, Heitsch, \& Mac Low 2000; 

479: Padoan \& Nordlund 2002; Vazguez-Semadeni, 

480: Ballesteros-Paredes, \& Klessen 2003; Shu \etal 2004;

481: Mac Low \& Klessen 2004).  

482: %Given the similarity of structure

483: %functions between molecular clouds determined in this study, 

484: %the role of compressible turbulence in generating such different 

485: %dense core conditions 

486: %would seem to be limited.

487: 

488: \section{Summary}

489: \parindent=20pt

490: We have examined the degree to which interstellar turbulence is universal 

491: within the molecular gas component of the Galaxy by comparing the 

492: measured structure functions for 27

493: giant molecular clouds.  Despite the large differences in cloud environments

494: and local star formation activity, the structure functions are 

495: described by very narrow ranges of the scaling exponent and scaling 

496: coefficient.  The degree of universality is further constrained by 

497: Monte Carlo simulations that replicate the observed scatter in the 

498: Larson scaling law that describes the relationship between global 

499: velocity dispersion and cloud size.  The near invariance of turbulence 

500: within the molecular interstellar medium suggests a common formation 

501: mechanism of molecular clouds such as shocks due to 

502: colliding gas streams of diffuse, 

503: atomic material as originally suggested by Larson (1981).  

504: It also implies that the energy contribution from 

505: internal sources such as stellar winds and expanding HII regions may be small 

506: with respect to a common, external component. 

507: 

508: \acknowledgments

509: We acknowledge useful comments from Richard Larson, John Scalo, Bruce

510: Elmegreen, Paolo Padoan, and Enrique Vazquez-Semadeni.

511: This work was supported by NSF grant AST 02-28993  to the Five College

512: Radio Astronomy Observatory.   

513: 

514: \begin{thebibliography}{}

515: 

516: \bibitem[Ballesteros-Paredes, Vazquez-Semadeni, \& Scalo 1999]{ballesteros99} 

517: Ballesteros-Paredes, J.E., Vazquez-Semadeni, E., \& Scalo, J. 1999, \apj,

518: 514, 286

519: \bibitem[Ballesteros \etal 1999]{b99} Ballesteros-Paredes, J.E., 

520: Hartmann, L., \& Vazquez-Semadeni, E. 1999, \apj, 527, 285

521: \bibitem[Ballesteros \& Mac Low]{b02 } Ballesteros-Paredes, J.E. 

522: \& Mac Low, M. 2002, \apj, 570, 734

523: \bibitem[Benson \& Myers 1989]{bm89} Benson, P.J. \& Myers, P.C. 1989, 

524: \apjs, 71, 89

525: \bibitem[Brunt \& Heyer 2002]{brunt02} Brunt, C.M. \& Heyer, M.H. 2002, 

526: \apj, 566, 276

527: \bibitem[Brunt 2003]{brunt03} Brunt, C.M. 2003, \apj, 584, 293

528: \bibitem[Brunt \etal 2003]{bruntetal04} Brunt, C.M., Heyer, M.H., 

529: Vazquez-Semadeni, E., \& Pichardo, B. 2003, \apj, 595, 824

530: \bibitem[Brunt \& Mac Low 2004]{bruntmaclow04} Brunt, C.M. \& Mac Low, M.

531: 2004, \apj, 604, 196

532: \bibitem[Brunt \etal 2004]{brunt04} Brunt, C.M., Heyer, M.H., Zivkov, V., 

533: \& Mac Low, M. 2004, \apj, in press

534: \bibitem[Carr 1987]{carr87} Carr, J.S. 1987, \apj, 323, 170

535: \bibitem[Elmegreen 1993]{elm93} Elmegreen, B.G. 1993, \apj, 419, L29

536: \bibitem[Falgarone \etal 1992]{falg92} Falgarone, E. Puget, J.-L., Pereault,

537: M. 1992, A\&A, 257, 715

538: \bibitem[Fuller \& Myers 1992]{fm92} Fuller, G.A. \& Myers, P.C. 1992, \apj, 384, 523

539: \bibitem[Heyer \& Schloerb 1997]{heyer97} Heyer, M.H. \& Schloerb, F.P. 1997, \apj, 475, 173

540: \bibitem[Heyer \etal 1998]{heyer98} Heyer, M.H., Brunt, C., Snell, R.L., 

541: Howe, J.E., Schloerb, F.P., \& Carpenter, J.M. 1998, \apjs, 115, 241

542: \bibitem[Hunter \etal 1986]{hunter86} Hunter, J.H., Sandford, M.T., Whitaker, R.W., \& Klein, R.I. 1986, \apj, 305, 309

543: \bibitem[Larson 1981]{larson81} Larson, R.B. 1981, \mnras,  194, 809

544: \bibitem[Loren 1989]{loren89} Loren, R.B., 1989, \apj, 338, 925

545: \bibitem[Klessen \etal 2000]{klessen00} Klessen, R.S., Heitsch, F., \& Mac Low, M. 2000, \apj, 535, 887

546: \bibitem[Mac Low \& Klessen 2004]{maclow02} Mac Low, M., \& Klessen, R.S.

547: 2004, {\it Reviews of Modern Physics}, 76, 125

548: \bibitem[Padoan \& Nordlund 2002]{pn02} Padoan, P. \& Nordlund, A. 2002, 

549: \apj, 576, 870

550: \bibitem[Pirogov etal 2003]{pirogov03} Pirogov, L., 

551: Zinchenko, I., Caselli, P., Johansson, L.E.B., \& Myers, P.C. 2003, A\&A, 

552: 405, 639

553: 

554: \bibitem[Shu \etal 2004]{shu04} Shu, F.H., Li, Z-Y, \& Allen, A. 2004, \apj, 601, 930

555: \bibitem[Simon \etal 2001]{simon01} Simon, R., Jackson, J.M., Clemens, D.P., 

556: Bania, T.M.,

557: \& Heyer, M.H. 2001, \apj, 551, 747

558: \bibitem[Solomon \etal 1987]{solomon87} Solomon, P.M., Rivolo, A.R., Barrett, J., \& Yahil, A. 1987, \apj, 319, 730 

559: 

560: \bibitem[Stutzki \& Gusten 1990]{stutzki90} St\"utzki, J. \& G\"usten, R.

561: 1990, \apj, 356, 513

562: \bibitem[Vazquez \etal 2003]{vazquez03} Vazquez, E., Ballesteros-Paredes, J.E. 

563: \& Klessen, R.S. 2003, \apj, 585, L131

564: \end{thebibliography}{}

565: 

566: \clearpage

567: 

568: \begin{figure}

569: \plotone {f1.eps}

570: \caption {The composite ${\delta}v,l$ relationship 

571: from PCA decompositions of \co\ J=1-0 imaging observations

572: of 27 individual molecular clouds.

573: The small scatter of points attest to the 

574: near invariance of interstellar turbulence within molecular clouds 

575: that exhibit a large range in size, environment, and star formation activity.

576: The large filled circles are the global velocity dispersion and size for each 

577: cloud derived from the first principal component. 

578: These are equivalent to the global velocity dispersion 

579: and size of the cloud as would be measured in the cloud-to-cloud size-line 

580: width relationship (Larson 1981; Solomon \etal 1987).  The light solid line 

581: show the bisector fit to all points from all clouds.  

582: The heavy solid line shows the

583: bisector fit to the filled circles exclusively.   The similarity of these 

584: two power laws explains the connection of Larson's cloud-to-cloud scaling 

585: law to the structure functions of individual clouds.

586: }

587: \label{composite}

588: \end{figure}

589: 

590: \begin{figure}

591: \plotone {f2.eps}

592: \caption

593: {

594: Images of \sigmasqr\ from the size-line width fits for an 

595: ensemble of clouds with varying turbulent properties 

596: parameterized by 

597: $\sigma_v$/$<v_\circ>$ and $\sigma_\gamma$/$<\gamma>$ for

598: different distances uncertainties ($\sigma_D/D$=0,0.1,0.25, and 0.5). 

599: The light contour values of \sigmasqr\ 

600: are 0.5, 1.0, 1.5 2, 2.5, 3.0 $km^2 s^{-2}$.

601: The heavy contour in each plot shows the loci that 

602: correspond to $\sigma_{obs}^2$=0.88 km$^2$ s$^{-2}$ 

603: from Solomon \etal (1987) and provides

604: an observational upper limit to cloud-to-cloud variations of 

605: $v_\circ$ and $\gamma$.  These loci

606: demonstrate that the structure function parameters, $\gamma$ and $v_\circ$,

607: can not 

608: vary by more than $\sim$10-20\%.

609: }

610: \label{mc_model}

611: \end{figure}

612: \end{document}

613: