1:
2: \section{Introduction}
3:
4: Fluid and plasma flows exhibit complex random behaviour at high Reynolds
5: number; this phenomena is called turbulence. On the Earth this phenomena
6: is seen in atmosphere, channel and rivers flows etc. In the universe,
7: most of the astrophysical systems are turbulent. Some of the examples
8: are solar wind, convective zone in stars, galactic plasma, accretion
9: disk etc.
10:
11: Reynolds number, defined as $UL/\nu$ ($U$ is the large-scale velocity,
12: $L$ is the large length scale, and $\nu$ is the kinematic viscosity),
13: has to be large (typically 2000 or more) for turbulence to set in.
14: At large Reynolds number, there are many active modes which are nonlinearly
15: coupled. These modes show random behaviour along with rich structures
16: and long-range correlations. Presence of large number of modes and
17: long-range correlations makes turbulence a very difficult problem
18: that remains largely unsolved for more than hundred years.
19:
20: Fortunately, random motion and presence of large number of modes make
21: turbulence amenable to statistical analysis. Notice that the energy
22: supplied at large-scales $(L)$ gets dissipated at small scales, say
23: $l_{d}$. Experiments and numerical simulations show that the velocity
24: difference $u(\mathbf{x+l})-u(\mathbf{x})$ has a universal probability
25: density function (pdf) for $l_{d}\ll l\ll L$. That is, the pdf is
26: independent of experimental conditions, forcing and dissipative mechanisms
27: etc. Because of its universal behaviour, the above quantity has been
28: of major interest among physicists for last sixty years. Unfortunately,
29: we do not yet know how to derive the form of this pdf from the first
30: principle, but some of the moments have been computed analytically.
31: The range of scales $l$ satisfying $l_{d}\ll l\ll L$ is called inertial
32: range.
33:
34: In 1941 Kolmogorov \cite{K41a,K41b,K41c} computed an exact expression
35: for the third moment of velocity difference. He showed that under
36: vanishing viscosity, third moment for velocity difference for homogeneous,
37: isotropic, incompressible, and steady-state fluid turbulence is\[
38: \left\langle \left(u_{||}(\mathbf{x+l})-u_{||}(\mathbf{x})\right)^{3}\right\rangle =\frac{4}{5}\Pi l\]
39: where $||$ is the parallel component along $\mathbf{l}$, $\left\langle .\right\rangle $
40: stands for ensemble average, and $\Pi$ is the energy cascade rate,
41: which is also equal to the energy supply rate at large scale $L$
42: and dissipation rate at the small scale $l_{d}$. Assuming fractal
43: structure for the velocity field, and $\Pi$ to be constant for all
44: $l$, we can show that the energy spectrum $E(k)$ is\[
45: E(k)=K_{Ko}\Pi^{2/3}k^{-5/3},\]
46: where $K_{Ko}$ is a universal constant, called Kolmogorov's constant,
47: and $L^{-1}\ll k\ll l_{d}^{-1}$. Numerical simulations and experiments
48: verify the above energy spectrum apart from a small deviation called
49: intermittency correction.
50:
51: Physics of magnetohydrodynamic (MHD) turbulence is more complex than
52: fluid turbulence. There are two coupled vector fields, velocity $\mathbf{u}$
53: and magnetic $\mathbf{b}$, and two dissipative parameters, viscosity
54: and resistivity. In addition, we have mean magnetic field $B_{0}$
55: which cannot be transformed away (unlike mean velocity field which
56: can be transformed away using Galilean transformation). The mean magnetic
57: field makes the turbulence anisotropic, further complicating the problem.
58: Availability of powerful computers and sophisticated theoretical tools
59: have helped us understand several aspects of MHD turbulence. In the
60: last ten years, there has been major advances in the understanding
61: of energy spectra and fluxes of MHD turbulence. Some of these theories
62: have been motivated by Kolmogorov's theory for fluid turbulence. Note
63: that incompressible turbulence is better understood than compressible
64: turbulence. Therefore, our discussion on MHD turbulence is primarily
65: for incompressible plasma. \emph{In this paper we focus on the universal
66: statistical properties of MHD turbulence, which are valid in the inertial
67: range. In this paper we will review the statistical properties of
68: following quantities:}
69:
70: \begin{enumerate}
71: \item Inertial-range energy spectrum for MHD turbulence.
72: \item Various energy fluxes in MHD turbulence.
73: \item Energy transfers between various wavenumber shells.
74: \item Anisotropic effects of mean magnetic field.
75: \item Structure functions $<\left(u_{||}(\mathbf{x+l})-u_{||}(\mathbf{x})\right)^{n}>$
76: and $<\left(b_{||}(\mathbf{x+l})-b_{||}(\mathbf{x})\right)^{n}>$,
77: where $u_{||}$ and $b_{||}$ are components of velocity and magnetic
78: fields along vector $\mathbf{l}$.
79: \item Growth of magnetic field (dynamo).
80: \end{enumerate}
81: Currently energy spectra and fluxes of isotropic MHD turbulence is
82: quite well established, but anisotropy, intermittency, and dynamo
83: is not yet fully understood. Therefore, items 1-3 will be discussed
84: in greater detail.
85:
86: Basic modes of incompressible MHD are Alfvén waves, which travel parallel
87: and antiparallel to the mean magnetic field with speed $B_{0}$. The
88: nonlinear terms induce interactions among these modes. In mid sixties
89: Kraichnan \cite{Krai:65} and Iroshnikov \cite{Iros} postulated that
90: the time-scale for the nonlinear interaction is proportional to $B_{0}^{-1}$,
91: leading to $E(k)\sim B_{0}^{1/2}k^{-3/2}$. However, research in last
92: ten years \cite{Srid2,MKV:B0_RG,Maro:Simulation,ChoVish:localB} show
93: that the energy spectrum of MHD turbulence Kolmogorov-like ($k^{-5/3}$).
94: Current understanding is that Alfvén waves are scattered by {}``local
95: mean magnetic field'' $B_{0}(k)\sim k^{-1/3}$, leading to Kolmogorov's
96: spectrum for MHD turbulence. The above ideas will be discussed in
97: Sections \ref{sec:Renormalization-Group-Analysis-MHD} and \ref{sec:Field-Theor-of-anisotropic}.
98:
99: In MHD turbulence there are exchanges of energies among the velocity-velocity,
100: velocity-magnetic, and magnetic-magnetic modes. These exchanges lead
101: to energy fluxes from inside of the velocity/magnetic wavenumber sphere
102: to the outside of the velocity/magnetic wavenumber sphere. Similarly
103: we have shell-to-shell energy transfers in MHD turbulence. We have
104: developed a new formalism called {}``mode-to-mode'' energy transfer
105: rates, using which we have computed energy fluxes and shell-to-shell
106: energy transfers numerically and analytically \cite{Dar:flux,MKV:MHD_PRE,MKV:MHD_Flux,MKV:MHD_Helical}.
107: The analytic calculations are based on field-theoretic techniques.
108: Note that some of the fluxes and shell-to-shell energy transfers are
109: possible only using {}``mode-to-mode'' energy transfer, and cannot
110: be computed using {}``combined energy transfer'' in a triad \cite{Lesi:book}.
111:
112: Many analytic calculations in fluid and MHD turbulence have been done
113: using field-theoretic techniques. Even though these methods are plagued
114: with some inconsistencies, we get many meaningful results using them.
115: In Sections \ref{sec:Renormalization-Group-Analysis-MHD}, \ref{sec:analytic-energy},
116: and \ref{sec:Field-Theor-of-anisotropic} we will review the field-theoretic
117: calculations of energy spectrum, energy fluxes, and shell-to-shell
118: energy transfers.
119:
120: Growth of magnetic field in MHD turbulence (dynamo) is of central
121: importance in MHD turbulence research. Earlier dynamo models (kinematic)
122: assumed a given form of velocity field and computed the growth of
123: large-scale magnetic field. These models do not take into account
124: the back-reaction of magnetic field on the velocity field. In last
125: ten years, many dynamic dynamo simulations have been done which also
126: include the above mentioned back reaction. Role of magnetic helicity
127: ($\mathbf{a}\cdot\mathbf{b}$, where $\mathbf{a}$ is the vector potential)
128: in the growth of large-scale magnetic field is better understood now.
129: Recently, Field et al. \cite{Fiel:Dynamo}, Chou \cite{Chou:theo},
130: Schekochihin et al. \cite{Maro:DynamoNlin} and Blackman \cite{Blac:Rev_Dynamo}
131: have constructed theoretical dynamical models of dynamo, and studied
132: nonlinear evolution and saturation mechanisms.
133:
134: As mentioned above, pdf of velocity difference in fluid turbulence
135: is still unsolved. We know from experiments and simulation that pdf
136: is close to Gaussian for small $\delta u$, but is nongaussian for
137: large $\delta u$. This phenomena is called intermittency. Note that
138: various moments called Structure functions are connected to pdf. It
139: can be shown that the structure functions are related to the {}``local
140: energy cascade rate'' $\Pi(k)$. Some phenomenological models, notably
141: by She and Leveque \cite{SheLeve} based on log-Poisson process, have
142: been developed to compute $\Pi(k)$; these models quite successfully
143: capture intermittency in both fluid and MHD turbulence. The predictions
144: of these models are in good agreement with numerical results. We will
145: discuss these issues in Section \ref{sec:Intermittency-in-MHD}.
146:
147: Numerical simulations have provided many important data and clues
148: for understanding the dynamics of turbulence. They have motivated
149: new models, and have verified/rejected existing models. In that sense,
150: they have become another type of experiment, hence commonly termed
151: as numerical experiments. Modern computers have made reasonably high
152: resolution simulations possible. The highest resolution simulation
153: in fluid turbulence is on $4096^{3}$ grid (e.g., by Gotoh \cite{Goto:DNS}),
154: and in MHD turbulence is on $1024^{3}$ grid (e.g., by Haugen et al.
155: \cite{Bran:Nonhelical_simulation}). Simulations of Biskamp \cite{Bisk:Kolm1,Bisk:Kolm2},
156: Cho et al. \cite{ChoVish:localB}, Maron and Goldreich \cite{Maro:Simulation}
157: have verified 5/3 spectrum for MHD turbulence. Dar et al. \cite{Dar:flux}
158: have computed various energy fluxes in 2D MHD turbulence. Earlier,
159: based on energy fluxes Verma et al. \cite{MKV:MHD_Simulation} could
160: conclude that Kolmogorov-like phenomenology models MHD turbulence
161: better that Kraichnan and Iroshnikov's phenomenology. Many interesting
162: simulations have been done to simulate dynamo, e. g., Chou \cite{Chou:num}
163: and Brandenburg \cite{Bran:Alpha}.
164:
165: Because of large values of dissipative parameters, MHD turbulence
166: requires large length and velocity scales. This make terrestrial experiments
167: on MHD turbulence impossible. However, astrophysical plasmas are typically
168: turbulent because of large length and velocity scales. Taking advantage
169: of this fact, large amount of solar-wind in-situ data have been collected
170: by spacecrafts. These data have been very useful in understanding
171: the physics of MHD turbulence. In fact, in 1982 Matthaeus and Goldstein
172: \cite{MattGold} had shown that solar wind data favors Kolmogorov's
173: $k^{-5/3}$ spectrum over Kraichnan and Iroshnikov's $k^{-3/2}$ spectrum.
174: Solar wind data also shows that MHD turbulence exhibits intermittency.
175: some of the observational results of solar wind will discussed in
176: Section \ref{sec:Solar-Wind}. In addition to the above topics, we
177: will also state the current results on the absolute equilibrium theories,
178: decay of global quantities, two-dimensional turbulence, shell model
179: of MHD turbulence, compressible turbulence etc.
180:
181: Literature on MHD turbulence is quite extensive. Recent book {}``Magnetohydrodynamic
182: Turbulence'' by Biskamp \cite{BiskTurb:book} covers most of the
183: basics. MHD turbulence normally figures as one of the chapters in
184: many books on Magnetohydrodynamics, namely Biskamp \cite{BiskNonl:book},
185: Priest \cite{Prie:book}, Raichoudhury \cite{Arna:book}, Shu \cite{Shu:book},
186: Cowling \cite{Cowl:book}, Vedenov \cite{Vede:book}. The recent developments
187: are nicely covered by the review articles in an edited volume \cite{Pass:book}.
188: Some of the important review articles are by Montgomery \cite{Mont:SW},
189: Pouquet \cite{Pouq:Rev}, Krommes \cite{Krom:Rev_Intermittency,Krom:Rev_Plasma}.
190: On dynamo, the key references are books by Moffatt \cite{Moff:book}
191: and Krause and Rädler \cite{Krau:book}, and recent review articles
192: \cite{Gilb:inbook,Robe:Rev_Geo,Bran:PR}. Relatively, fluid turbulence
193: has a larger volume of literature. Here we will list only some of
194: the relevant ones. Leslie \cite{Lesl:book}, McComb \cite{McCo:book,McCo:rev,McCo:RGBook},
195: Zhou et al. \cite{ZhouMcCo:RGrev}, and Smith and Woodruff \cite{Smit}
196: have reviewed field-theoretic treatment of fluid turbulence. The recent
197: books by Frisch \cite{Fris:book} and Lesieur \cite{Lesi:book} cover
198: recent developments and phenomenological theories. The review articles
199: by Orszag \cite{Orsz:Rev}, Kraichnan and Montgomery \cite{KraiMont},
200: and Sreenivasan \cite{Sree:RMP} are quite exhaustive.
201:
202: In this review paper, we have focussed on statistical theory of MHD
203: turbulence, specially on energy spectra, energy fluxes, and shell-to-shell
204: energy transfers. These quantities have been analyzed analytically
205: and numerically. A significant portion of the paper is devoted to
206: self-consistent field-theoretic calculations of MHD turbulence and
207: {}``mode-to-mode'' energy transfer rates because of their power
208: of analysis as well as our familiarity with these topics. These topics
209: are new and are of current interest. Hence, this review article complements
210: the earlier work. Universal laws are observed in the inertial range
211: of homogeneous and isotropic turbulence. Following the similar approach,
212: in analytic calculations of MHD turbulence, homogeneity and isotropy
213: are assumed except in the presence of mean magnetic field.
214:
215: To keep our discussion focussed, we have left out many important topics
216: like coherent structures, astrophysical objects like accretion disks
217: and Sun, transition to turbulence etc. Our discussion on compressible
218: turbulence and intermittency is relatively brief because final word
219: on these topics still awaited. Dynamo theory is only touched upon;
220: the reader is referred to the above mentioned references for a detailed
221: discussion. In the discussion on the solar wind, only a small number
222: of results connected to energy spectra are covered.
223:
224: The outline of the paper is as follows: Section \ref{sec:MHD:-Definitions-and}
225: contains definition of various global and spectral quantities along
226: with their governing equations. In Section \ref{sec:Mode-to-mode-Energy-Transfer}
227: we discuss the formalism of {}``mode-to-mode'' energy transfer rates
228: in fluid and MHD turbulence. Using this formalism, formulas for energy
229: fluxes and shell-to-shell energy transfer rates have been derived.
230: Section \ref{sec:MHD-Turbulence-Models} contains the existing MHD
231: turbulence phenomenologies which include Kraichnan's 3/2 mode; Kolmogorov-like
232: models of Goldreich and Sridhar. Absolute equilibrium theories and
233: Selective decay are also discussed here. In Section \ref{sec:Solar-Wind}
234: we review the observed energy spectra of the solar wind. Section \ref{sec:Numerical-Investigation-MHD}
235: describes Pseudo-spectral method along with the numerical results
236: on energy spectra, fluxes, and shell-to-shell energy transfers. In
237: these discussions we verify which of the turbulence phenomenologies
238: are in agreement with the solar wind data and numerical results.
239:
240: Next three sections cover applications of field-theoretic techniques
241: to MHD turbulence. In Section \ref{sec:Renormalization-Group-Analysis-MHD}
242: we introduce Renormalization-group analysis of MHD turbulence, with
243: an emphasis on the renormalization of {}``mean magnetic field''
244: \cite{MKV:B0_RG}, viscosity and resistivity \cite{MKV:MHD_RG}. In
245: Section \ref{sec:analytic-energy} we compute various energy fluxes
246: and shell-to-shell energy transfers in MHD turbulence using field-theoretic
247: techniques. Here we also review eddy-damped quasi-normal Markovian
248: (EDQNM) calculations of MHD turbulence. In Section \ref{sec:Field-Theor-of-anisotropic}
249: we discuss the anisotropic turbulence calculations of Goldreich and
250: Sridhar \cite{Srid1,Srid2} and Galtiers et al. \cite{Galt:Weak}
251: in significant details. The variations of turbulence properties with
252: space dimensions have been discussed.
253:
254: In Section \ref{sec:Magnetic-Field-Growth} we briefly mention the
255: main numerical and analytic results on homogeneous and isotropic dynamo.
256: We include both kinematic and dynamic dynamo models, with emphasis
257: being on the later. Section \ref{sec:Intermittency-in-MHD} contains
258: a brief discussion on intermittency models of fluid and MHD turbulence.
259: Next section \ref{sec:Miscellaneous-Topics} contains a brief discussion
260: on the large-eddy simulations, decay of global energy, compressible
261: turbulence, and shell model of MHD turbulence. Appendix \ref{sec:Fourier-Series}
262: contains the definitions of Fourier series and transforms of fields
263: in homogeneous turbulence. Appendix B and C contain the Feynman diagrams
264: for MHD turbulence; these diagrams are used in the field-theoretic
265: calculations. In the last Appendix \ref{sec:Digression-to-Fluid}
266: we briefly mention the main results of spectral theory of fluid turbulence
267: in 2D and 3D.
268:
269:
270: