1: \documentclass[12pt,preprint]{aastex}
2: % \documentclass[manuscript]{aastex}
3: % \documentclass[preprint2]{aastex}
4: % \documentclass{emulateapj}
5: % \documentclass[onecolumn]{emulateapj}
6: \usepackage{epsf}
7:
8: % \slugcomment{\today}
9:
10: \shorttitle{DENSITY POWER SPECTRUM}
11: \shortauthors{KIM \& RYU}
12:
13: \begin{document}
14:
15: \title{DENSITY POWER SPECTRUM OF COMPRESSIBLE HYDRODYNAMIC TURBULENT
16: FLOWS}
17:
18: \author{Jongsoo Kim}
19: \affil{Korea Astronomy and Space Science Institute, Hwaam-Dong,
20: Yuseong-Gu, Daejeon 305-348, Korea: {\tt jskim@kasi.re.kr}}
21:
22: \and
23:
24: \author{Dongsu Ryu}
25:
26: \affil{Department of Astronomy and Space Science, Chungman National
27: University, Daejeon 305-764, Korea: {\tt ryu@canopus.cnu.ac.kr}}
28:
29: \begin{abstract}
30:
31: Turbulent flows are ubiquitous in astrophysical environments, and
32: understanding density structures and their statistics in turbulent
33: media is of great importance in astrophysics.
34: In this paper, we study the density power spectra, $P_{\rho}$, of
35: transonic and supersonic turbulent flows through one and
36: three-dimensional simulations of driven, isothermal hydrodynamic
37: turbulence with root-mean-square Mach number in the range of
38: $1 \la M_{\rm rms} \la 10$.
39: From one-dimensional experiments we find that the slope of the
40: density power spectra becomes gradually shallower as the rms Mach
41: number increases.
42: It is because the density distribution transforms from the profile
43: with {\it discontinuities} having $P_{\rho} \propto k^{-2}$ for
44: $M_{\rm rms} \sim 1$ to the profile with {\it peaks} having
45: $P_{\rho} \propto k^0$ for $M_{\rm rms} \gg 1$.
46: We also find that the same trend is carried to three-dimension; that is,
47: the density power spectrum flattens as the Mach number increases.
48: But the density power spectrum of the flow with $M_{\rm rms} \sim 1$
49: has the Kolmogorov slope.
50: The flattening is the consequence of the dominant density structures
51: of {\it filaments} and {\it sheets}.
52: Observations have claimed different slopes of density power spectra
53: for electron density and cold H I gas in the interstellar medium.
54: We argue that while the Kolmogorov spectrum for electron density
55: reflects the {\it transonic} turbulence of $M_{\rm rms} \sim 1$
56: in the warm ionized medium, the shallower spectrum of cold H I gas
57: reflects the {\it supersonic} turbulence of $M_{\rm rms} \sim$ a few
58: in the cold neutral medium.
59:
60: \end{abstract}
61:
62: \keywords{hydrodynamics --- methods:numerical --- turbulence}
63:
64: \section{INTRODUCTION}
65:
66: According to the currently accepted paradigm, the interstellar medium
67: (ISM) is in a state of turbulence and the turbulence is believed to
68: play an important role in shaping complex structures of velocity
69: and density distributions \citep[see, e.g.,][for reviews]{vopg00,es04}.
70: The ISM turbulence is transonic or supersonic with Mach number varying
71: from place to place; while the turbulent Mach number, $M$, is probably
72: of order unity in the warm ionized medium (WIM) judging from the
73: temperature that ranges from $6 \times 10^3$ to $10^4$ K \citep{hrt99},
74: it is a few in the cold neutral medium (CNM) \citep[e.g.,][]{ht03}
75: and as large as 10 or higher in molecular clouds \citep[e.g.,][]{lars81}.
76: The fact that $M \ga 1$ implies high compressibility,
77: and observations of the ISM reflect the density structures resulted
78: from the turbulence.
79: Interpreting these observations and hence understanding the density
80: structures require a good knowledge of density statistics.
81:
82: Density power spectrum, $P_{\rho}$, is a statistics that can be
83: directly extracted from observations.
84: The best-known density power spectrum of the ISM is the one
85: presented in \citet{acr81,ars95}.
86: It is a composite power spectrum of {\it electron density} collected
87: from observations of velocity fluctuations of the interstellar gas,
88: rotation measures, dispersion measures, interstellar scintillations
89: and others.
90: What is remarkable is that a single Kolmogorov slope of $-5/3$ fits
91: the power spectrum for the wavenumbers spanning more than 10 decades.
92: But there are also a number of observations that indicate
93: shallower spectra.
94: For instance, \citet{ddg00} showed that the density power spectrum
95: of {\it cold H I gas} has a much shallower slope of $-0.75 \sim -0.5$.
96: Here we note that the power spectrum of \citet{acr81,ars95} should
97: reveal density fluctuations preferably in the WIM since it is for
98: the electron density.
99: On the other hand, the power spectrum of \citet{ddg00} represents the
100: so-called tiny-scale atomic structure in the CNM \citep{heil97}.
101:
102: Discussions on density power spectrum in the context of theoretical
103: works have been scarce.
104: It is partly because the study of turbulence was initiated in the
105: incompressible limit.
106: Instead velocity power spectrum was discussed extensively by
107: comparing the well-known theoretical spectra, such as those of
108: \citet{kolm41} and \citet{gs95}.
109: Hence, although hydrodynamic or magnetohydrodynamic (MHD)
110: simulations of compressible turbulence were performed,
111: the spectral analysis was focused mainly on velocity
112: (and magnetic field in MHDs) \citep[e.g.,][]{cl02,vos03}.
113: A few recent simulation studies, however, have reported density
114: power spectrum \citep[e.g.,][]{pjjn04,knp04,blc05}.
115: For instance, in an effort to constrain the average magnetic field
116: strength in molecular cloud complexes, \citet{pjjn04} have
117: simulated MHD turbulence of root-mean-square sonic Mach number
118: $M_{\rm rms} = 10$ with weak and equipartition magnetic fields.
119: They have found that the slope of density power spectrum is steeper
120: in the weak, super-Alfv\'enic case than in the equipartition case,
121: and super-Alfv\'enic turbulence may be more consistent with
122: observations.
123: Although their work has focused on the dependence of the slope of
124: density power spectrum on Alfv\'enic Mach number, they have also
125: demonstrated that the slope is much shallower than the Kolmogorov
126: slope.
127: \citet{knp04} and \citet{blc05} have found consistently shallow
128: slopes.
129: However, none of the above studies have yet systematically
130: investigated the dependency of the slope of density power spectrum
131: on sonic Mach number.
132:
133: A note worthwhile to make is that the functional form of probability
134: distribution function (PDF) on the {\it density} field of compressible
135: turbulent flows is well established.
136: \citet{pv98}, by performing one-dimensional isothermal hydrodynamic
137: simulations, have shown that the density PDF follows a log-normal
138: distribution.
139: A few groups \citep[e.g.,][]{np99,osg01} have reported that the
140: log-normal distribution still holds in three-dimensional isothermal
141: turbulent flows even though the scaling of the standard deviation
142: with respect to rms Mach number is not necessarily the same as
143: the one obtained by \citet{pv98}.
144:
145: In this paper we study the density distribution and the density
146: power spectrum in transonic and supersonic turbulent flows using
147: one and three-dimensional hydrodynamic simulations.
148: Specifically we show that the Mach number is an important parameter
149: that characterizes the density power spectrum.
150: The plan of the paper is as follows.
151: In \S 2 a brief description on numerical details is given.
152: Results are presented in \S 3, followed by summary and discussion
153: in \S 4.
154:
155: \section{SIMULATIONS}
156:
157: Isothermal, compressible hydrodynamic turbulence was simulated using
158: a code based on the total variation diminishing scheme \citep{krjh99}.
159: Starting from an initially uniform medium with density $\rho_o$ and
160: isothermal sound speed $a$ in a box of size $L$,
161: the turbulence was driven, following the usual procedure
162: \citep[e.g.,][]{macl99}; velocity perturbations were added at every
163: $\Delta t =0.001 L/a$, which were drawn from a Gaussian random field
164: determined by the top-hat power distribution in a narrow wavenumber
165: range of $1 \leq k \leq 2$.
166: The dimensionless wavenumber is defined as $k \equiv L/\lambda$, which
167: counts the number of waves with wavelength $\lambda$ inside $L$.
168: The amplitude of perturbations was fixed in such a way that $M_{\rm rms}$
169: of the resulting flows ranges approximately from 1 to 10.
170: Self-gravity was ignored.
171:
172: One and three-dimensional numerical simulations were performed with
173: 8196 and $512^3$ grid zones, respectively.
174: We note that one-dimensional turbulence has compressible mode only
175: but no rotational mode, and so its application to real situations is
176: limited.
177: However, there are two advantages: 1) high-resolution can be
178: achieved, resulting in a wide inertial range of power spectra and
179: 2) the structures formed can be easily visualized and understood.
180: In the three-dimensional experiments the driving of turbulence was
181: enforced to be incompressible by removing the compressible part in
182: the Fourier space, although our results are not sensitive to the
183: details of driving.
184: Table 1 summarizes the model parameters.
185:
186: \section{RESULTS}
187:
188: Figure 1 presents the results of one-dimensional simulations.
189: Top two panels are snapshots of the spatial profiles of velocity
190: and density for a transonic turbulence with $M_{\rm rms} = 0.8$
191: (left) and a supersonic turbulence $M_{\rm rms} = 12.5$ (right).
192: Shocks are developed in both cases, but weak shocks in the transonic
193: case and strong shocks in the supersonic case.
194: The velocity profile shows shock discontinuities, superimposed
195: on the background that reminisces the $k=1$ and $2$ driving.
196: Especially, the supersonic case exhibits the so-called {\it sawtooth
197: profile}, which is expected in the turbulence dominated by shocks.
198: Like the velocity profile, the density profile for the transonic case
199: shows shock {\it discontinuities}.
200: But the density profile for the supersonic case shows high {\it peaks},
201: which can be understood as follows.
202: Since the density jump is proportional to the square of Mach number
203: in isothermal shocks, the density contrast at high Mach number shocks
204: is very large.
205: With the total mass conserved, the fluid should be concentrated at
206: shock discontinuities producing a peak-dominated profile.
207:
208: Bottom two panels of Figure 1 show the time-averaged power spectra
209: of velocity (left) and density (right) for flows with different
210: $M_{\rm rms}$'s (see Table 1 for the time interval for averaging).
211: We note that the statistical errors (or standard deviations) of the
212: spectra, which are not drawn in the figure, easily exceed the averaged
213: values, themselves.
214: This is partially due to the fact that the statistical sample is small
215: in one-dimensional experiments.
216: The slope of the velocity power spectra is nearly equal to $-2$,
217: irrespective of rms Mach numbers, as shown in the bottom-left panel.
218: On the contrary, while the slope of the density power spectrum is
219: close to $\sim -2$ for transonic and mildly supersonic flows, it is
220: much shallower for supersonic flows with $M_{\rm rms} = 7.5$ and 12.5.
221: Such power spectrum reflects the profiles in the top panels.
222: We note that discontinuities (i.e., step functions) and infinitely
223: thin peaks (i.e., delta-functions) in spatial distribution result in
224: $P_{\rho} \propto k^{-2}$ and $P_{\rho} \propto k^0$, respectively.
225: Hence while the density profiles with discontinuities for transonic
226: flows give power spectra with slopes close to $-2$, the profiles with
227: peaks for highly supersonic flows give shallower spectra.
228:
229: Interestingly $P_{\rho} \propto k^{-2}$ and $P_{\rho} \propto k^0$
230: were predicted in the context of the Burgers turbulence.
231: It is known that the Burgers equation describes a one-dimensional
232: turbulence with randomly developed shocks, and the resulting sawtooth
233: profile gives to the $k^{-2}$ velocity power spectrum, i.e.,
234: $P_v \propto k^{-2}$ \citep[see, e.g.,][]{ff83}.
235: Expanding it, \citet{sw96} developed a model for the description
236: of density advected in a velocity field governed by the Burgers
237: equation.
238: They showed that in the limit of negligible pressure force (i.e.,
239: for strong shocks), all the mass concentrates at shock discontinuities
240: and the density power spectrum becomes $P_{\rho} \propto k^0$.
241: In the presence of pressure force (i.e., for weak shocks) their density
242: power spectrum is $P_{\rho} \propto k^{-2}$.
243: Our results agree with those of \citet{sw96}, although ours are
244: based on numerical simulations using full hydrodynamic equations
245: (under the assumption of isothermal flows).
246:
247: The flattening of density power spectrum for supersonic turbulence is
248: also seen in three-dimensional experiments.
249: The results are presented in Figure 2.
250: Top and middle panels show the density distribution in a two-dimensional
251: slice for a transonic turbulence with $M_{\rm rms} = 1.2$ (top) and
252: a supersonic turbulence with $M_{\rm rms} = 12$ (middle).
253: The images reveal different morphologies for the two cases.
254: The transonic image includes curves of {\it discontinuities}, which
255: are surfaces of shocks with density jump of a few.
256: So the three-dimensional density distribution should contain
257: surfaces of discontinuities on the top of smooth turbulent background.
258: On the other hand, the supersonic image shows mostly density
259: {\it concentrations} of string and dot shapes, which are sheets and
260: filaments in three-dimension.
261: Hence, the three-dimensional density distribution should be dominated
262: by sheets and filaments of high density concentration.
263:
264: Bottom panel of Figure 2 shows the time-averaged density power
265: spectra for flows with different $M_{\rm rms}$'s, along with fitted
266: values of slope (see Table 1 for the time interval for averaging).
267: As in one-dimensional experiments, the slope over the inertial range
268: becomes shallow as the Mach number increases.
269: But unlike in one-dimension, the slope for the transonic turbulence
270: with $M_{\rm rms} = 1.2$ is $-1.73$, close to the Kolmogorov value
271: of $-5/3$.
272: It is because the power spectrum represents the fluctuations of
273: turbulent background, rather than discontinuities of weak shocks.
274: Note that in three-dimension turbulence has the rotational mode
275: of eddy motions in addition to the compressional mode of sound
276: waves and shock discontinuities, and the normal cascade is allowed.
277: In supersonic turbulence, we get the slopes of $-1.08$, $-0.75$
278: and $-0.52$ for $M_{\rm rms} = 3.4$, $7.3$ and $12.0$, respectively.
279: Again the shallow spectrum reflects the highly concentrated density
280: distribution, but for this time, of sheet and filament morphology
281: shown in the middle panel.
282: Like delta-functions, infinitely thin sheets and filaments in
283: three-dimension give $P_{\rho} \propto k^0$.
284:
285: As pointed in Introduction, a few recent studies of turbulence
286: have published the slope of density power spectrum: for instance,
287: \citet{pjjn04} have found a slope of $-0.71$ for $M_{\rm rms} = 10$
288: MHD turbulence with weak magnetic field, and \citet{knp04} have found
289: a slope of $-0.88$ for $M_{\rm rms} = 6$ hydrodynamic turbulence.
290: We notes that these values are well consistent with ours.
291:
292: \section{SUMMARY AND DISCUSSION}
293:
294: We present the density distribution and the density power spectrum
295: of driven, isothermal, compressible hydrodynamic turbulence in one
296: and three-dimensional numerical experiments.
297: The rms Mach number of flows covers from $\sim 1$ (transonic) to
298: $\sim 10$ (highly supersonic).
299: Our main findings are summarized as follows.\\
300: 1. In transonic turbulence, the density distribution is characterized
301: by {\it discontinuities} generated by weak shocks on the top of
302: turbulent background.
303: On the other hand, in supersonic turbulence with $M_{\rm rms} \gg 1$,
304: it is characterized by {\it peaks} or {\it concentrations} of mass
305: generated by strong shocks.
306: Those density concentrations appear as sheets and filaments in
307: three-dimension.\\
308: 2. In three-dimension, the slope of density power spectrum for
309: transonic turbulence with $M_{\rm rms} = 1.2$ is $-1.73$, which is
310: close to the Kolmogorov slope of $-5/3$.
311: But as the rms Mach number increases, the slope flattens, reflecting
312: the development of sheet-like and filamentary density structures.
313: The slopes of supersonic turbulence with $M_{\rm rms} = 3.4$, $7.3$
314: and $12.0$ are $-1.08$, $-0.75$ and $-0.52$, respectively.
315:
316: In Introduction, we have pointed that in the ISM the power spectrum
317: of electron density has the Kolmogorov slope \citep{acr81,ars95},
318: whereas the power spectrum of H I gas shows a shallower slope
319: \citep{ddg00}.
320: Our results suggest the reconciliation of these claims of different
321: spectral slopes.
322: That is, the electron density power spectrum represents the density
323: fluctuations mostly in the WIM, where the turbulence is transonic and
324: the density power spectrum has the Kolmogorov slope.
325: On the other hand, the H I power spectrum represents the tiny-scale
326: atomic structures in the CNM, where the turbulence is supersonic with
327: $M_{\rm rms} \sim$ a few and the density power spectrum has a shallower
328: slope.
329:
330: Through 21cm H I observations, it has been pointed that sheets
331: and filaments could be the dominated density structure in the CNM
332: \citep[e.g.,][]{heil97,ht03}.
333: Our study demonstrates that sheets and filaments are indeed the
334: natural morphological structures in the media with supersonic
335: turbulence such as the CNM.
336:
337: In this work we demonstrate that the slope of density power spectrum
338: becomes shallow as the rms Mach number increases by considering the
339: simplest possible physics, i.e., by neglecting magnetic field and
340: self-gravity and assuming isothermality.
341: We point, however, that those physics could affect the
342: {\it quantitative} results.
343: For instance, the magnetic field, which is dynamically important
344: in the ISM, could further shallow the density power spectrum,
345: as \citet{pjjn04} have shown.
346: In addition, self-gravity could affect the power spectrum significantly,
347: since it forms clumps and causes the density power spectrum to become
348: flatter or even to have positive slopes.
349: Finally, cooling could make a difference too, especially in the
350: turbulence in the WIM, and may even enhance compressibility,
351: although, as a first order approximation, the assumption of isothermality
352: would be adequate in molecular clouds \citep[see, e.g.,][]{psm05}.
353:
354: \acknowledgments
355:
356: We thank the referee, P. Padoan, for constructive comments, and
357: J. Cho and H. Kang for discussions.
358: The work by JK was supported by KOSEF through Astrophysical
359: Research Center for the Structure and Evolution of Cosmos (ARCSEC).
360: The work by DR was supported by Korea Research Foundation Grant
361: (KRF-2004-015-C00213). Numerical simulations were performed using
362: ``Linux Cluster for Astronomical Computations'' of KASI-ARCSEC.
363:
364: \begin{thebibliography}{}
365:
366: \bibitem[Armstrong et al.(1981)]{acr81}
367: Armstrong, J. W., Cordes, J. M. \& Rickett, B. J. 1981, \nat, 291, 561
368:
369: \bibitem[Armstrong et al.(1995)]{ars95}
370: Armstrong, J. W., Rickett, B. J. \& Spangler, S. R. 1995, \apj, 443, 209
371:
372: \bibitem[Beresnyak et al.(2005)]{blc05}
373: Beresnyak, A., Lazarian, A. \& Cho, J. 2005, \apj, 624, L93
374:
375: \bibitem[Cho \& Lazarian(2002)]{cl02}
376: Cho, J. \& Lazarian, A. 2002, \prl, 88, 245001
377:
378: \bibitem[Deshpande et al.(2000)]{ddg00}
379: Deshpande, A. A., Dwarakanath, K. S. \& Goss, W. M. 2000, \apj, 543, 227
380:
381: \bibitem[Elmegreen \& Scalo(2004)]{es04}
382: Elmegreen, B. G. \& Scalo, J. 2004, \araa, 42, 211
383:
384: \bibitem[Fournier \& Frisch(1983)]{ff83}
385: Fournier, J. D. \& Frisch, U. 1983, J. Mec. Theor. Appl., 2, 699
386:
387: \bibitem[Goldreich \& Sridhar(1995)]{gs95}
388: Goldreich, P. \& Sridhar, H. \apj, 438, 763
389:
390: \bibitem[Haffner et al.(1999)]{hrt99}
391: Haffner, L. M., Reynolds, R. J. \& Tufte, S. L. 1999, \apj, 523, 223
392:
393: \bibitem[Heiles(1997)]{heil97}
394: Heiles, C. 1997, \apj, 481, 193
395:
396: \bibitem[Heiles \& Troland(2003)]{ht03}
397: Heiles, C. \& Troland, T. H. 2003, \apj, 586, 1067
398:
399: \bibitem[Kim et al.(1999)]{krjh99}
400: Kim, J., Ryu, D., Jones, T. W. \& Hong, S. S. 1999, \apj, 514, 506
401:
402: \bibitem[Kolmogorov(1941)]{kolm41}
403: Kolmogorov, A. 1941, Dokl. Akad. Nauk SSSR, 31, 538
404:
405: \bibitem[Kritsuk et al.(2004)]{knp04}
406: Kritsuk, A., Norman, M. \& Padoan, P. 2004 (astro-ph0411626)
407:
408: \bibitem[Larson (1981)]{lars81}
409: Larson, R. B. 1981, \mnras, 194, 809
410:
411: \bibitem[Mac Low(1999)]{macl99}
412: Mac Low, M.-M. 1999, \apj, 524, 169
413:
414: \bibitem[Nordlund \& Padoan(1999)]{np99}
415: Nordlund, \AA. \& Padoan, P. 1999, in Interstellar Turbulence, eds.
416: J. Franco \& A. Carrami\~nana (Cambridge: Cambridge Univ. Press), 218
417:
418: \bibitem[Ostriker et al.(2001)]{osg01}
419: Ostriker, E. C., Stone, J. M. \& Gammie, C. F. 2001, \apj, 546, 980
420:
421: \bibitem[Pavlovski et al.(2005)]{psm05}
422: Pavlovski, G., Smith, M. D. \& Mac Low, M.-M. 2005 (astro-ph/0504504)
423:
424: \bibitem[Padoan et al.(2004)]{pjjn04}
425: Padoan, P., Jimenez, R., Juvela, M. \& Nordlund, \AA. 2004, \apj, 604, L49
426:
427: \bibitem[Passot \& V\'azquez-Semadeni(1998)]{pv98}
428: Passot, T. \& V\'azquez-Semadeni, E. 1998, \pre, 58, 4501
429:
430: \bibitem[Saichev \& Woyczynski(1996)]{sw96}
431: Saichev, A. I. \& Woyczynski, W. A. 1996, SIAM J. Appl. Math., 56, 1008
432:
433: \bibitem[V\'azquez-Semadeni et al.(2000)]{vopg00}
434: V\'azquez-Semadeni, E., Ostriker, E. C., Passot, T., Gammie, C. F.
435: \& Stone, J. M. 2000, in Protostars and Planets IV, eds. V, Mannings,
436: A. P. Boss \& S. S. Russell (Tuscon: University of Arizona Press), p 3
437:
438: \bibitem[Vestuto et al.(2003)]{vos03}
439: Vestuto, J. G., Ostriker, E. C. \& Stone, J. M. 2003, \apj, 590, 858
440:
441: \end{thebibliography}
442:
443: \clearpage
444:
445: \begin{deluxetable}{lcllcc}
446: \tabletypesize{\scriptsize}
447: \tablecaption{Model Parameters\tablenotemark{a}}
448: \tablewidth{0pt}
449: \tablehead{
450: \colhead{Model\tablenotemark{b}} &
451: \colhead{$\stackrel{\cdot}{E}_{\rm kin}$\tablenotemark{c}} &
452: \colhead{$t_{\rm end}$\tablenotemark{d}} &
453: \colhead{$dt_{\rm output}$\tablenotemark{e}} &
454: \colhead{$\Delta t_{\rm sp}$\tablenotemark{f}} &
455: \colhead{Resolution}
456: }
457: \startdata
458: 1D0.8 & 0.1 & 8 & 0.1 & 4-8 & 8196 \\
459: 1D1.7 & 1 & 8 & 0.1 & 4-8 & 8196 \\
460: 1D3.4 & 10 & 8 & 0.1 & 4-8 & 8196 \\
461: 1D7.5 & 100 & 8 & 0.1 & 4-8 & 8196 \\
462: 1D12.5 & 400 & 8 & 0.1 & 4-8 & 8196 \\
463: 3D1.2 & 1 & 4.8 & 0.4 & 1.2-4.8 & $512^3$ \\
464: 3D3.4 & 30 & 1.5 & 0.1 & 0.6-1.5 & $512^3$ \\
465: 3D7.3 & 300 & 1.0 & 0.05 & 0.5-1.0 & $512^3$ \\
466: 3D12.0 & 1300& 0.5 & 0.05 & 0.3-0.5 & $512^3$ \\
467: \enddata
468: \tablenotetext{a}{All the quantities are given in the units of
469: $\rho_o$, $a$ and $L$.}
470: \tablenotetext{b}{1D or 3D for one or three-dimension followed by
471: the rms Mach number.}
472: \tablenotetext{c}{kinetic energy input rate}
473: \tablenotetext{d}{end time of each simulation}
474: \tablenotetext{e}{time interval for data output}
475: \tablenotetext{f}{time interval over which power spectra
476: were averaged}
477: \end{deluxetable}
478:
479: \clearpage
480:
481: \begin{figure}
482: \vskip -2cm \plotone{f1.eps} \vskip -5cm \figcaption
483: {One-dimensional turbulence.
484: {\it Top panels}: Snapshots of the spatial profiles of velocity
485: and density, after turbulence was fully saturated, for a transonic
486: turbulence with time averaged $M_{\rm rms} = 0.8$ (left) and a
487: supersonic turbulence with time averaged $M_{\rm rms} = 12.5$ (right).
488: The plotted quantities are normalized by the isothermal sound speed
489: and the initial density, respectively.
490: {\it Bottom panels}: Time-averaged power spectra of velocity (left) and
491: density (right) for flows with different $M_{\rm rms}$'s.
492: The velocity power spectrum in the left panel is multiplied by $k^2$
493: for clarity.
494: Slopes in the right panel were obtained by least-square fits over
495: the range of $20 \le k \le 60$.
496: Simulations used 8196 grids zones.}
497:
498: \end{figure}
499:
500: \begin{figure}
501: \vspace{0cm}\hspace{4.5cm}\epsfxsize=6.5cm\epsfbox{f2.eps}\vspace{0cm}
502: \figcaption
503: {Three-dimensional turbulence.
504: {\it Top and middle panels}: Color images of the density distribution
505: in a two-dimensional slice, after turbulence was fully saturated, for a
506: transonic turbulence with time averaged $M_{\rm rms} = 1.2$ (top) and
507: a supersonic turbulence with time averaged $M_{\rm rms} = 12$ (middle).
508: The color is coded on linear scales and the range of density is
509: an order of magnitude larger for the supersonic case than for the
510: transonic case.
511: {\it Bottom panel}: Statistical error bars of the time-averaged density
512: power spectra for flows with different $M_{\rm rms}$'s.
513: Solid lines and their slopes, which were obtained by least-square fits
514: over the range of $4 \le k \le 14$, are included.
515: Simulations used $512^3$ grids zones.}
516: \end{figure}
517:
518: \end{document}
519: