1: \documentclass[12pt]{article}

2: \usepackage{amsmath}

3: \usepackage{psfig}

4: \setlength{\oddsidemargin}{-.05in} 

5: \setlength{\evensidemargin}{00in}

6: \setlength{\textwidth}{6.50in}

7: \setlength{\topmargin}{-.5in}

8: \setlength{\textheight}{8.50in}

9: \renewcommand{\theequation}{\thesection .\arabic{equation}}

10: \renewcommand{\baselinestretch}{1.4}

11: \newcommand{\bfchi}{\mbox{\boldmath $\chi$}}

12: \newcommand{\bfxi}{\mbox{\boldmath $\xi$}}

13: %\newtheorem{Theorem}{Theorem}[section]

14: %\newenvironment{Proof}{P{\sc roof. }}{$\Box$}

15: %\newtheorem{Corollary}{Corollary}[section]

16: %\newtheorem{Lemma}{Lemma}[section]

17: %\def\mybibliography#1{\list{}{\setlength{\leftmargin}{2em}\setlength{\labelsep}{0%pt}\itemindent=-\leftmargin}

18: %\def\newblock{\hskip .11em plus .33 em minus -.07em}\sloppy\clubpenalty4000\widow%penalty4000\sfcode`\.=1000\relax}

19: \newcommand{\bfphi}{\mbox{\boldmath $\phi$}}

20: \newcommand{\bfomega}{\mbox{\boldmath $\omega$}}

21: %\setlength{\parskip}{2mm}

22: %\pagestyle{empty}

23: 

24: \begin{document}

25: 

26: \title{Assessing Turbulence Strength via Lyaponuv Exponents}

27:                    

28: \author{Mayer Humi\\

29: Department of Mathematical Sciences\\

30: Worcester Polytechnic Institute\\

31: 100 Institute Road\\

32: Worcester, MA  01609}

33: 

34: \maketitle

35: \thispagestyle{empty}

36: 

37: \begin{abstract}

38: In this paper we study the link between `turbulence strength' in a flow

39: and the leading Lyaponuv exponent that characterize it. To this end we

40: use two approaches. The first, analytical, considers the truncated

41: convection equations in 2-dimensions with three (Lorenz model) and six

42: components and study their leading Lyaponuv exponent as a function of the

43: Rayleigh number. For the second approach we analyze fifteen time series 

44: of measurements taken by a plane flying at constant height in the upper 

45: troposphere. For each of these time series we estimate the leading Lyaponuv 

46: exponent which we then correlate with the structure constant for the 

47: temperature.

48: \end{abstract}

49: 

50: \newpage

51: 

52: 

53: {\bf Turbulence is still considered as one of the major unsolved problems of

54: modern science. Usually one attempts to characterize the turbulent state 

55: of the system in terms of various numbers such as the Reynolds number,

56: Grashof number, Rayleigh number, etc. How a combination of these numbers 

57: actually characterize the strength of turbulence in the system is

58: not answered easily. The situation is even more complex when the flow is

59: represented only by a time series. Furthermore is there any relation 

60: between this `strength' and chaos theory?  Although chaos owes it modern 

61: origins to the work of Lorenz who investigated the onset of convection, 

62: a characterization of turbulence in terms of dynamical invariants such

63: as Lyapunov exponents, fractal dimension and so on is still an open

64: question which needs further investigation. This paper attempts to 

65: present a modest contribution in this direction.}

66: 

67: \newpage

68: 

69: \section{Introduction}

70: 

71: Atmospheric convection in two dimensions is described [1,2] by the following 

72: system of partial differential equation:

73: \begin{equation}

74: \label{1}

75: \frac{\partial}{\partial t} \nabla^2\psi = -

76: \frac{\partial(\psi,\nabla^2\psi)}{\partial(x,z)} + \nu \nabla^4\psi +

77: \alpha g \frac{\partial\theta}{\partial x}

78: \end{equation}

79: \begin{equation}

80: \label{2}

81: \frac{\partial\theta}{\partial t} =

82: -\frac{\partial(\psi,\theta)}{\partial(x,z)} + \frac{\Delta T}{H}

83:  \;\frac{\partial\psi}{\partial x} + \kappa\nabla^2\theta

84: \end{equation}

85: 

86: Subject to the stress free boundary conditions:

87: 

88: \begin{equation}

89: \psi=0,\,\,\nabla^2\psi=0,\,\,\theta=0 ,\,\,\,\, z=0,1

90: \end{equation}

91: In this system $\psi$ is the stream function and $\theta$ is the potential

92: temperature. The constants $\alpha, \nu, \kappa, g$ denote respectively

93: the coefficient of thermal expansion,the kinematic viscosity,the thermal

94: conductivity and the acceleration of gravity. $H$ is the fluid layer 

95: thickness

96: and $\Delta T$ is the temperature difference between the upper and lower

97: surface of the fluid (which is assumed to be held constant). We also have

98: \begin{equation}

99: \label{3}

100: \frac{\partial(f,g)}{\partial(x,z)} = \left| \begin{array}{cc}

101: \frac{\partial f}{\partial x}    &\frac{\partial f}{\partial z}\\

102: \frac{\partial g}{\partial x}   &\frac{\partial g}{\partial z}

103: \end{array}

104: \right|

105: \end{equation}

106: 

107: 

108: In his famous 1963 paper [2] Lorenz introduced and studied a model in which

109: the solution to eq.(\ref{1})-(\ref{2}) is approximated by a three Fourier modes

110: \begin{equation}

111: \label{4}

112: \psi = \frac{\kappa(1+a^2)\sqrt{2}}{a}X(t)\sin\left(\frac{\pi

113: ax)}{H}\right) \sin\left(\frac{\pi z}{H}\right)

114: \end{equation}

115: \begin{equation}

116: \label{5}

117: \theta = \frac{R_c}{\pi R_a}\left\{\sqrt{2} Y(t) \cos\left(\frac{\pi a

118: x}{H}\right)\sin\left(\frac{\pi z}{H}\right) -

119: Z(t)\sin\left(\frac{2\pi z}{H}\right)\right\}

120: \end{equation}

121: where $a$ is a parameter and

122: \begin{equation}

123: R_a = \frac{g \alpha H^3 \Delta T}{\kappa \nu}

124: \end{equation}

125: \begin{equation}

126: R_c = \frac{\pi^4 (1+a^2)^3}{a^2}

127: \end{equation}

128: are the Rayleigh number and the critical Rayleigh numbers for the flow.

129: This led to the following three coupled equations for $X,Y,Z$.

130: \begin{eqnarray}

131: \label{6}

132: \dot{X} &=& -\sigma X + \sigma Y\\ \nonumber

133: \dot{Y} &=& XZ + rX - Y\\ \nonumber

134: \dot{Z} &=& XY - bZ

135: \end{eqnarray}

136: where

137: \begin{equation}

138: \label{7}

139: \sigma = \frac{\nu}{\kappa},\,\, b = \frac{4}{(1 + a^2)}, \,\, 

140: r = \frac{R_a}{R_c}.

141: \end{equation}

142: 

143:    

144: These equations are usually referred to as the `Lorenz model'. Since its

145: appearance 

146: this model and its implications were studied in great detail in hundreds of 

147: publications with special attention to its bifurcations as a function 

148: of the parameters $\sigma, r$ and $b$. It has been recognized [3] 

149: however that (as expected) the approximation of the solution to eqs 

150: (\ref{1}) and (\ref{2}) which is provided by (\ref{3})-(\ref{7}) becomes 

151: `poor' as $r$ increases.  This has 

152: led several authors to develop and study models with a larger number of 

153: modes [4,21]. 

154: 

155: The basic conjecture that we want to validate in this paper, through some

156: case study, is that stronger turbulence [22] in a system is linked to larger 

157: value of the leading Lyaponuv exponents [5,6] for the the flow. 

158: To test this conjecture we

159: study the Lyaponuv exponents of two truncated systems that were derived form 

160: eqs (\ref{1})-(\ref{2}). We then use Rosenstein algorithm [7] to

161: estimate the 

162: Lyaponuv exponents for fifteen atmospheric time series that were obtained

163: from measurements in the upper troposphere (10Km approximately).

164: The Lyaponuv exponent is then correlated with the structure constant 

165: for the temperature [8,9] which is representative of the 

166: density fluctuations

167: in the atmosphere and hence with the strength of turbulence in the flow.

168: 

169: To begin with we study the dependence of the leading

170: Lyaponuv exponent on the parameter $r$ in the Lorenz model. This parameter

171: is representative of the Rayleigh number in eqs. (1.1)-(1.2).

172: We show that for small values of $r$ the leading Lyaponuv exponent in 

173: the Lorenz model is either negative or small. Increasing

174: $r$ leads at the `onset of chaos' to a steep increase in the value of

175: this exponent. Further increase in $r$ leads to further modest increase 

176: in the value of the leading Lyaponuv exponent (which is indicative of the

177: fact that these modes are 'saturated').

178: 

179: One expects to obtain a better approximation to the solution of eqs.

180: (\ref{1})-(\ref{2}) as the number of modes allowed in the model increases. 

181: Hence as a second step in our study we consider a model with six modes.

182: The leading Lyaponuv exponent for this model displays a similar behavior

183: as for the Lorenz model. However the highest value of the leading Lyaponuv 

184: exponent in this model is larger than the one attained by the Lorenz model. 

185: This is consistent with our conjecture since a model with a larger number 

186: of modes can represent stronger turbulent state of the system due 

187: to the larger number of interacting modes. 

188: 

189: From a practical point of view it is important to assess turbulence

190: strength in a flow from a representative time series of measurements.

191: In this case although the equations governing the flow are well known

192: one has no initial or boundary conditions to simulate them. Furthermore

193: it is impossible to estimate correctly the parameters that appear in these 

194: equations 

195: from the data at hand. To gauge the turbulence strength under these

196: circumstances one has to proceed indirectly. One well known manifestation

197: of turbulence strength in the atmosphere is through density fluctuations

198: (this leads to the well known `twinkling of the stars' phenomena).

199: Hence turbulence strength is related to the value $C_N^2$ (the refraction

200: structure constant [8,9]). In the upper troposphere (at

201: heights of about 10km)

202: the main contributor to $C_N^2$ is $C_T^2$ -the structure constant for the 

203: temperature. It follows then that if our conjecture is correct then

204: the leading Lyaponuv exponent for the temperature time series should 

205: increase as $C_T^2$ increases.

206: 

207: The plan of the paper is as follows:\,\, In section 2 we introduce the 

208: six mode truncated model for eqs (\ref{1}) -(\ref{2}) and study the 

209: behaviors of 

210: the leading Lyaponuv exponent for the Lorenz model and this system as a 

211: function of $r$. In section 3 we give a short overview about the atmospheric

212: structure constants and their computation. In section 4 we consider the time 

213: series of measurements for the temperature and show that there exist a 

214: strong correlation between $C_T^2$ and the leading Lyaponuv exponent. 

215: We end up in section 5 with summary and conclusions.

216: 

217: \setcounter{equation}{0}

218: \section{Analytical Approach}

219: 

220: In an attempt to relate the leading Lyaponuv exponent  to

221: the turbulence strength in the flow we consider a truncated expansion 

222: of eqs.  (\ref{1})-(\ref{2}) with 3 and 6 Fourier components. For the six 

223: components model we let 

224: \begin{eqnarray}

225: \label{21}

226: \psi &=&

227: \frac{\sqrt{2}(1+a^2)\kappa}{a}\left\{\left[X_1(t)\sin\left(\frac{\pi

228: ax}{H}\right) + X_2(t)\sin\left(\frac{2\pi

229: ax}{H}\right)\right]\right.\\ \nonumber

230: &&\sin\left(\frac{\pi z}{H}\right)\left.  +

231: X_3(t)\sin\left(\frac{\pi ax}{H}\right)\sin\left(\frac{2\pi

232: z}{H}\right)\right\}

233: \end{eqnarray}

234: \begin{equation}

235: \label{2.2}

236: \theta = \frac{\sqrt{2}\Delta T R_c}{\pi R_a}

237: \left[Y_1(t)\cos\left(\frac{\pi ax}{H}\right) + Y_2(t)\cos\left(\frac{2\pi

238: ax}{H}\right)\right]\sin\left(\frac{\pi z}{H}\right) - \frac{\Delta

239: T R_a}{\pi R_c} Z(t)\sin\left(\frac{2\pi z}{H}\right)

240: \end{equation}

241: which leads after some lengthy algebra to the following six equations

242: for ${X_1,X_2,X_3,Y_1,Y_2,Z}$

243: \begin{equation}

244: \label{23}

245: \dot{X}_1 = \sigma Y_1 - \sigma

246: X_1+\frac{9}{4}\frac{(a^2-1)\sqrt{2}}{1 + a^2} X_2X_3

247: \end{equation}

248: \begin{equation}

249: \label{24}

250: \dot{X}_2 = \frac{2\sigma(1+a^2)}{1+4a^2}Y_2 - \frac{\sigma(1+4a^2)}{1+a^2}

251: X_2 + \frac{9\sqrt{2}}{4(1+4a^2)}\;X_1X_3

252: \end{equation}

253: \begin{equation}

254: \label{25}

255: \dot{X}_3 = -\frac{\sigma(4+a^2)}{1+a^2}\;X_3 -

256: \frac{9a^2\sqrt{2}}{4(4+a^2)} X_1X_2

257: \end{equation}

258: \begin{equation}

259: \label{26}

260: \dot{Y}_1 = \frac{3}{4}\sqrt{2} X_3Y_2 - X_1 Z - Y_1 + rX_1

261: \end{equation}

262: \begin{equation}

263: \label{27}

264: \dot{Y}_2 = -\frac{3}{4}\sqrt{2} X_3Y_1 - \frac{(1+4a^2)}{1+a^2}Y_2 -

265: 2X_2Z + 2rX_2

266: \end{equation}

267: \begin{equation}

268: \label{28}

269: \dot{Z} = X_1Y_1 + 2X_2Y_2 - bZ

270: \end{equation}

271: 

272: 

273: For the systems (\ref{5})-(\ref{7}) and (\ref{21})-(\ref{28}) we applied

274: (the analytical) Wolf 

275: algorithm [10] to compute the leading Lyaponuv exponent as a function

276: of $r$. The results of these computations are shown in Fig. 1 for

277: values of $r$ in the range of [10,115]. We observe that both of these systems 

278: represent a truncation of the original equations. As a result the flow 

279: which is represented by eqs.(\ref{1})-(\ref{2}) can not be approximated well by 

280: the solution of the truncated equations when $r$ is large as additional modes 

281: become active. As a result the relationship between the solutions of eqs.

282: (\ref{5})-(\ref{7}) and (\ref{23})-(\ref{28}) and actual convective turbulence 

283: phenomena is lost as $r$ becomes large. 

284: 

285: From Fig. 1 we see that for both truncated systems under consideration

286: the leading Lyaponuv exponent remain small or negative for small values

287: of $r$. It then ``bifurcates'' and goes up sharply at the onset of convective 

288: state and increases modestly as $r$ is increased further. For the

289: three mode model as $r$ increases over 70 the modes become saturated 

290: and the model ceases to approximate the true solution of the convective 

291: problem. It then follows it own characteristics. 

292: 

293: For the six mode solution, the leading Lyaponuv exponent 

294: follows the same trajectory as the three mode model up

295: to $r ~ 70$. However for higher values of $r$ the additional modes in this

296: model become active and as a result the leading Lyaponuv exponent continue 

297: to increase up to $r ~ 110$ (where the modes of this model get saturated).

298: 

299: The obvious explanation of this behavior of the leading Lyaponuv exponent

300: is that the  

301: the six mode system provides a better approximation to the solution of eqs.

302: (\ref{1})-(\ref{2}) than the three mode system. As a result the six mode system

303: captures more of the turbulence phenomena which are related to the solution

304: of the original system. Accordingly Fig. 1 confirms our conjecture and shows

305: that the flow in a system with stronger turbulence has also a larger

306: leading Lyaponuv exponent. 

307: 

308: \setcounter{equation}{0}

309: \section{Structure Constants}

310: 

311: The structure function of a geophysical variable e.g. the temperature is

312: defined as

313: 

314: \begin{equation}

315: D({\bf r})= \langle T^\prime[({\bf r}_1 + {\bf r}),T^\prime ({\bf r}_1)]^{2}\rangle

316: \end{equation}

317: 

318: where $T^\prime$ are the turbulent fluctuations in the temperature and 

319: ${\bf r}$ is the vector from one point to another.

320: 

321: Kolmogorov showed that for isotropic turbulence in the inertial range this

322: function depends only on $d$ = $|{\bf r}|$ and scales as

323: 

324: \begin{equation}

325: D(d) = C_T^{2} d^{2/3} .

326: \end{equation}

327: 

328: $C_T^2$ which appears as the proportionality constant in this equation

329: is referred to as the ``temperature structure constant''.

330: 

331: The determination of the atmospheric structure constants [8,9,11,12]

332: and in particular the temperature structure constant $C_T^2$ is important

333: in many applications e.g the propagation of electromagnetic signals

334: [8,9]. Local peaks in the values of these constants, which are indicative 

335: of strong turbulence and reflect on the structure of the atmospheric flow, 

336: can have a negative effect on the operation of various optical instruments.

337: 

338: To estimate these structure constants in the upper troposphere or the

339: stratosphere it is a common practice to send high flying airplanes

340: that collect data about the basic meteorological variables (such as wind,

341: temperature and pressure) along its flight path which may extend up

342: to 200 kms. To estimate the averaged value of the structure constants

343: along this path one must decompose first the meteorological data into

344: mean flow, waves and turbulent residuals [13,14]. From the spectrum of the

345: turbulent residuals one can estimate an averaged value of the structure

346: constants using Kolmogorov inertial range scaling and Taylor's frozen

347: turbulence hypothesis [15]. For $C_T^2$ in particular we have

348: \begin{equation}

349:             C_T^2(k) = 4 F(k) k^{5/3}

350: \end{equation}

351: where $F(k)$ is the temperature spectral density in the inertial range

352: and $k$ is the wave number. An averaged value for $C_T^2$ (over all wave

353: numbers in the inertial range) is obtained by averaging these

354: values over $k$.

355: 

356: In the upper troposphere where humidity is low $C_T^2$ is the main

357: contributor to the refraction structure constant $C_N^2$ [8,9] which has 

358: an important impact on the operation of ground telescopes and the 

359: twinkling of the stars.

360: 

361: \section{Analysis of Atmospheric Time Series}

362: 

363: During 1999 a series of measurements were made over Australia and Japan

364: by a specially equipped aeroplane [14]. These measurements were taken with 

365: sampling frequency of 55.1 Hz along a flight path of 200 kms (approximately). 

366: For each data set the plane flew at almost constant height along a 

367: straight line at an approximate speed of 103m/sec [14,15].

368: 

369: To use this data to estimate $C_T^2$ we have to split the original

370: measurements into a sum of mean flow, waves and turbulent residuals.

371: To accomplish this task we used Karahunen-Loeve(K-L) decomposition

372: algorithm which has been used by many researchers [15,16].

373: Actually the last paper applies the K-L decomposition to

374: the same data considered here and the details of the decomposition

375: which were given there will not be repeated here.

376: 

377: Based on instrument specifications the data noise should be at a relative

378: error level of $10^{-3}$. This is confirmed by the eigenvalues obtained

379: in the K-L decomposition where the last few eigenvalues (which reflect

380: the noise level in the data) are of order $10^{-3}$ of the leading 

381: eigenvalue.

382: 

383: For fifteen data sets that were obtained during these flights averaged 

384: values for $C_T^2$ were obtained using the methodology described above 

385: [8,9,10].

386: 

387: To compute the leading Lyaponuv exponents for these time series we used

388: Rosenstein algorithm and its implementation in the TISEAN package [7,17]. 

389: (However one should note that other algorithms are available for this 

390: purpose [18]).

391: To apply this algorithm one has to determine first the `optimal'

392: delay coordinates and embedding dimension. These were determined using the

393: mutual information and false neighborhood algorithms [5,19,20] which are 

394: also implemented in the above mentioned package. This analysis led us to 

395: choose a four dimensional embedding space with delay coordinates of 600

396: data points. Fig 2. presents a log-log plot of the results 

397: for this exponent versus $C_T^2$. We also show on this plot the least

398: square line for this data. From this plot we see that a change in $C_T^2$

399: over three orders of magnitudes correlates well with the leading Lyaponuv 

400: exponents. The fluctuations around the least squares line can be attributed

401: to wave activity and possible measurements errors. This demonstrates that

402: the Lyaponuv exponent can be used as a second global measure of turbulence

403: strength in the data. In cases of discrepancy between $C_T^2$ and 

404: the leading Lyaponuv exponent one must trace out the reasons for this 

405: mismatch and correct them.

406: 

407: \section{Summary and Conclusions}

408: 

409: Our main objective in this paper was to link turbulence strength to

410: the leading Lyaponuv exponent that is related to the flow. This program

411: was carried out in two contexts. In the first we used a truncated expansion 

412: of the flow and tested for leading Lyaponuv exponent of the resulting dynamical 

413: system. In the second part we tested this hypothesis for time series

414: which represent the flow. In both instances we found a strong supporting 

415: evidence for our conjecture. Although this does not constitutes a proof

416: of this conjecture in general we still feel that our results will

417: useful in many applied contexts especially as a check for the validity of

418: other global invariants that characterize the flow. In particular the 

419: determination of the leading Lyaponuv exponent can help verify the value

420: of $C_T^2$ (or other structure constants) that have important practical 

421: applications. 

422: 

423: 

424: \section*{References}

425: 

426: \begin{itemize}

427: 

428: \item[1] Saltzman, B., 1962:\, Finite amplitude free convection as an 

429: initial value problem, {\it J. Atmos. Sci}, {\bf 19}, 329-341.

430: 

431: \item[2] Lorenz, E.N., 1963:\,  Deterministic nonperiodic flow,

432: {\it J. Atmos. Sci}, {\bf 20}, 130-141.

433: 

434: \item[3] Marcus, P. S., 1981:\, Effects of truncation in modal representation

435: of thermal convection, {\it J. Fluid Mech.}, {\bf 103}, 241-255. 

436: 

437: \item[4] Curry, J. H., 1978:\, A generalized Lorenz system, 

438: {\it Commun. Math. Phys.}, {\bf 60}, 193-204.

439: 

440: \item[5] Abarbanel, H. D. I., Brown, R., Sidorowich, J.J., and Tsimring L.S.,

441: 1993:\, The analysis of observed chaotic data in physical systems,

442: {\it Rev. Mod. Phys.}, {\bf 65}, 1331.

443: 

444: \item[6] Ott, E., Sauer, T., Yorke, J. A., 1994:\, Coping with Chaos, 

445: Wiley, NY. 

446: 

447: \item[7] Rosenstein, M. T., Collins, J. J., and De Luca, C. J., 1993:\,

448: {\it Physica D}, {\bf 65}, 117.

449: 

450: \item[8] Panofsky, H.A., and Dutton, J.A., 1989:\,  Atmospheric

451: Turbulence,  Wiley, NY.

452: 

453: \item[9] Dewan, E., 1980: \, Optical Turbulence Forecasting,  {\it Air Force

454: Geophysical Lab.}, AFGL-TR-80-0030, Lab. Report.

455: 

456: \item[10] Fanta, R.L., 1985:\,  Wave Propagation in Random Media,  {\it

457: Progress in Optics}, {\bf 22}, 342-399.

458: 

459: \item[11] Wolf, A, Swift, J. B., Swinney, H.L., and Vastano, J.A., 1985:\;

460: Determining Lyapunov exponents from a time series, {\it Physica D},

461: {\bf 16}, 285-317.

462: 

463: \item[12] Humi, M., 2004:\; Estimation of Atmospheric Structure Constants

464: from Airplane Data, {\it J. of Atmospheric and Oceanic Technology},

465: {\bf 21}, 495-500. 

466: 

467: \item[13] Jumper, G.Y. and Beland, R.R., Progress in the understandings and

468: modeling of atmospheric optical turbulence, {\it AIAA}, 2000-2355,

469: Proc. 31st AIAA Plasmadynamics and Lasers Conference, 19-22 June 2000,

470: Denver, CO.

471: 

472: \item[14] Cote, O.R., Hacker J.M., Crawford, T.L., and Dobosy, R.J., 2000:\, 

473: Clear air turbulence and refractive turbulence in the upper troposphere

474: and lower stratosphere, Ninth conference on Aviation, Range and Aerospace

475: Meteorology, Paper No. 16386, Sep 2000.

476: 

477: \item[15] Humi, M., 2003:\, Approximate Simulation of Stratospheric Flow 

478: from Aircraft Data,

479: {\it Inter. J. Numerical Methods in Fluids}, {\bf 41}, 209-223.

480: 

481: \item[16] Penland, C., Ghil, M., and Weickmann, K.M., 1991:\, Adaptive filtering

482: and maximum entropy spectra with applications to changes in atmospheric

483: angular momentum, {\it J. Geo. Res.}, {\bf 96}, 22659-22671.

484: 

485: \item[17] Hegger, R,  Kantz, H., and Schreiber, T., 1991:\, Practical 

486: implementation of nonlinear time series methods:\, The TISEAN package, 

487: Chaos {\bf 9}, 413.

488: 

489: \item[18] Liu, H., Yang, Y., Dai, Z., Yu, Z., 2003:\, The largest Lyapunov

490: exponent of a chaotic dynamical system in scale space and its

491: application, {\it Chaos}, {\bf 13}, 839-844.

492: 

493: \item[19] Fraser, A.M. and Swinney, H.L., 1986:\, Independent coordinates for 

494: strange attractors from mutual information, Phys. Rev. A {\bf 33}, 1134.

495: 

496: \item[20] Kennel, M.B., Brown, R. and Abarbanel, H.D.I., 1992:\, Determining 

497: embedding dimension for phase-space reconstruction using a geometrical 

498: construction, Phys. Rev. A, {\bf 45}, 3403.

499: 

500: \item[21] Reiterer P.,Lainscsek C., Schurrer F., Letellier C. and Maquet J.

501: 1998: A nine dimensional Lorenz system to study high dimensional chaos

502: J. Phys. A {\bf 31} 7121-7139.

503: 

504: \item[22] Frisch U. and Orszag S. -Turbulence: Challenges for theory and 

505: experiment, Physics today {\bf 43} ,24-29 

506: 

507: \end{itemize}

508: 

509: \newpage

510: 

511: \centerline{\Large{\bf List of Captions}}

512: 

513: Fig. 1: Plot of the leading Lyapunov exponent versus $r=\frac{R_a}{R_c}$

514: for the Lorenz model(dashed line) and the 6-mode model (solid line)

515: which was developed in Sec.2.

516: 

517: Fig. 2: Log-log plot of E - the leading Lyapunov exponent- versus $C_T^2$. 

518: The first order least squares fit for this data yields the relation

519: $E=10.85*(C_T^2)^(0.2760)$ .

520: 

521: \newpage

522: 

523: \begin{figure}[htb]

524: \centering

525: \centerline{\psfig{file=fig01.ps,height=4in,width=5.5in}}

526: \caption{}

527: \end{figure}

528: 

529: \newpage

530: 

531: \begin{figure}[htb]

532: \centering

533: \centerline{\psfig{file=fig02.ps,height=4in,width=5.5in}}

534: \caption{}

535: \end{figure}

536: 

537: \end{document} 

538: