1: %\documentstyle[12pt,aasms4]{article}
2: %\documentstyle[12pt,aas2pp4]{article}
3: %\documentstyle[12pt,apjpt4]{article}
4:
5: %% manuscript produces a one-column, double-spaced document:
6:
7: %\documentclass[manuscript]{aastex}
8:
9: %% preprint2 produces a double-column, single-spaced document:
10:
11: % \documentclass[preprint2]{aastex}
12: \documentclass[12pt,preprint]{aastex}
13:
14: %\usepackage{amsfonts}
15: %\usepackage{amsmath}
16:
17: % Macros
18: \newcommand{\br}{\mathbf{r}}
19: \newcommand{\bV}{\mathbf{V}}
20: \newcommand{\bB}{\mathbf{B}}
21: \newcommand{\bE}{\mathbf{E}}
22: \newcommand{\bJ}{\mathbf{J}}
23: \newcommand{\rot}{\mathbf{\nabla} \times}
24:
25: \shorttitle{Solar wind small scale inertial range}
26: \shortauthors{Alexandrova et al.}
27:
28: \begin{document}
29: \title{Small scale energy cascade of the solar wind turbulence}
30: \author{O. Alexandrova}
31: \affil{LESIA/CNRS, Observatoire de Paris, 5 pl. J. Janssen, 92195
32: Meudon, France}
33: \author{V. Carbone, P. Veltri}
34: \affil{Dip.~Fisica/CNISM, Ponte P. Bucci, Cubo 31C, 87036 Rende
35: (CS), Italy}
36:
37: %Dipartimento di Fisica/CNISM -- Unit\`a di Cosenza, Ponte Pietro Bucci, Cubo 31C, 87036 Rende (CS), Italy
38:
39: \and
40: \author{L. Sorriso--Valvo}
41: \affil{LICRYL/CNR, Ponte P. Bucci, Cubo 33B, 87036 Rende (CS),
42: Italy}
43:
44: %%\date{\today}
45:
46: %\input epsf
47:
48: \begin{abstract}
49: Magnetic fluctuations in the solar wind are distributed according to
50: Kolmogorov's power law $f^{-5/3}$ below the ion cyclotron frequency
51: $f_{ci}$. Above this frequency, the observed steeper power law is
52: usually interpreted in two different ways: a dissipative range of
53: the solar wind turbulence or another turbulent cascade, the nature
54: of which is still an open question. Using the Cluster magnetic data
55: we show that after the spectral break the intermittency increases
56: toward higher frequencies, indicating the presence of non-linear
57: interactions inherent to a new inertial range and not to the
58: dissipative range. At the same time the level of compressible
59: fluctuations raises. We show that the energy transfer rate and
60: intermittency are sensitive to the level of compressibility of the
61: magnetic fluctuations within the small scale inertial range. We
62: conjecture that the time needed to establish this inertial range is
63: shorter than the eddy-turnover time, and is related to dispersive
64: effects. A simple phenomenological model, based on the compressible
65: Hall MHD, predicts the magnetic spectrum $\sim k^{-7/3+2\alpha}$,
66: which depends on the degree of plasma compression $\alpha$.
67: \end{abstract}
68: \keywords{solar wind turbulence, Hall MHD} \maketitle
69:
70:
71:
72: \section{Introduction} \label{Intro}
73:
74: Solar wind, which is highly turbulent, represents a unique
75: opportunity to investigate turbulence in natural plasmas via
76: \emph{in situ} measurements \cite{sw2,noi}. In non-magnetized
77: fluids, where the energy injection scale is far from the dissipation
78: one, the intermediate scales (inertial range) are described by the
79: universal power law Kolmogorov spectrum $k^{-s}$ with $s=5/3$. This
80: law depends neither on the energy injection nor on the energy
81: dissipation processes.
82:
83: In the solar wind, and in the interplanetary space in general, the
84: mean free path roughly corresponds to the Sun-Earth distance and the
85: usual dissipation via collisions is negligible. At the same time, in
86: a magnetized plasma there is a number of characteristic space and
87: temporal scales. Investigating solar wind turbulence at these scales
88: is then a challenging topic from the point of view of basic plasma
89: physics. Here we will focus our discussion on the ion scales, namely
90: the ion inertial length $\lambda_i = c/\omega_{pi}$ and the ion
91: cyclotron frequency $f_{ci}= eB/m_i$. At these scales the fluid-like
92: approximation of plasma dynamics, the usual Magnetohydrodynamic
93: (MHD) description, breaks down in favor of a more complex
94: description of plasma.
95:
96: Solar wind turbulent spectrum of magnetic field fluctuations follows
97: a $\sim f^{-5/3}$ power law below the ion cyclotron frequency
98: $f_{ci}$. For $f>f_{ci}$ the spectrum steepens significantly, but is
99: still described by a power law $f^{-s}$, with $s \in (2-4)$
100: \citep{Leamon98,smith06}.
101:
102: \emph{In situ} solar wind measurements provide time series data,
103: i.e. information on frequency in Fourier space. How to get any
104: information on wave vectors? For fluctuations with velocities much
105: smaller than the plasma bulk velocity $V$, the Taylor hypothesis is
106: valid: the observed variations on a time scale $\delta t$
107: correspond to variations on the spatial scale $\delta r=V \delta t$.
108: Therefore there is a direct correspondence between $f$ and $k$
109: spectra. %In the solar wind this is valid if we assume strong
110: %turbulence (without linear waves).
111: If the solar wind turbulence was a mixture of linear wave modes,
112: Taylor hypothesis would be well verified only below the spectral
113: break: the bulk velocity is superalfv\'enic ($V>V_A$, where $V_A$
114: being the Alfv\'en speed) and so the low frequency fluctuations can
115: be considered as frozen in plasma. However, whistler waves (with $f
116: > f_{ci}$ and phase speed $V_{\varphi} > V_A$) do not satisfy the
117: Taylor hypothesis assumption. In this study we assume that in the
118: solar wind there are no whistler waves above the spectral break
119: frequency. This last assumption is supported by results recently
120: obtained in the Earth's magnetosheath \citep{Mangeney2006}.
121:
122:
123: Using the Taylor hypothesis the observed solar wind spectrum below
124: the break is usually attributed to the Kolmogorov's spectrum $\sim
125: k^{-5/3}$. Above the spectral break, the spectral steepening, $\sim
126: k^{-3}$, can be interpreted in two different ways. Some authors
127: associate it to the dissipation range
128: \citep{Leamon98,Leamon99,Leamon2000,Bale2005,smith06}. Others
129: suggest that after the spectral break another turbulent cascade
130: takes place \citep{Biskamp1996,Ghosh96,stawicki,Li01,Galtier06}.
131:
132: In ordinary fluid flows the dissipation range is described by an
133: exponential function \citep{frisch}. While in the solar wind, a
134: well-defined power law is observed after the break point and not an
135: exponential. Note that power spectra, i.e. the second order
136: statistics, completely describe Gaussian, or statistically
137: independent, fluctuations. However, as is well known, fluctuations
138: cannot be described by a Gaussian statistics in the low-frequency
139: part of the solar wind turbulence \citep{noi}. Deviations from
140: Gaussianity, i.e. intermittency \citep{frisch}, may be quantified by
141: the flatness, the forth-order moment of fluctuations.
142:
143:
144: In this paper we investigate the nature of magnetic fluctuations in
145: the high frequency range of the solar wind turbulence, that is
146: usually called dissipation range. We find that in this range the
147: flatness increase with frequency. This is similar to what is going
148: on in the low-frequency range. The presence of intermittency
149: together with the well defined power law in the high-frequency part
150: of the spectrum suggests another turbulent cascade rather than a
151: dissipation range. This small scale cascade is observed to be much
152: more compressible than the Kolmogorov-like inertial range, in
153: agreement with the previous observations by the Wind spacecraft
154: \citep{Leamon98}. We show that the energy transfer rate and
155: intermittency are sensitive to the level of compressibility of the
156: turbulent fluctuations in this range. Finally, we propose a simple
157: phenomenological model, based on the compressible Hall MHD, which
158: allows to explain the observed range of the spectral indices in the
159: high frequency part of the solar wind spectrum.
160:
161:
162: \section{Turbulent spectra and intermittency}
163:
164: In the present study we analyze the Cluster magnetic field data up
165: to $12.5$ Hz (or $100f_{ci}$). We use 57 minutes of data during the
166: interval 22:35:00--23:32:00 UT on 5 April 2001, when Cluster was in
167: the solar wind not connected to the bow-shock. During this period
168: the interplanetary magnetic filed is $B\simeq 7$ nT, the ion density
169: is $n_i\simeq 3$ cm$^{-3}$, the plasma bulk velocity is $V\simeq
170: 540$ km/s, the ion temperature is $T_i \simeq 33$~eV, the Alfv\'en
171: speed is $V_A=B/\sqrt{\mu _0 m_in_i}\simeq 90$ km/s, the ion skin
172: depth $\lambda_i$ is about $130$~km, the ion Larmor radius is
173: $\rho_i \simeq 80$~km and the ion plasma beta is $\beta_i=2\mu _0
174: p_i/B^2\simeq 0.8$, $p_i$ being the pressure of the ions. All the
175: data are from Cluster spacecraft one, $B$ is determined from the
176: fluxgate magnetometer (FGM) \citep{Balogh2001}, the other plasma
177: parameters are provided by CIS/HIA instrument \citep{CIS2001}.
178: Magnetic field fluctuations are measured by the search coils (SC) of
179: the STAFF experiment with $0.04$~s time resolution
180: \citep{Cornilleau2003}. This instrument\footnote{STAFF-SC measurs
181: magnetic fluctuations above 0.1~Hz. As the Cluster spin frequency is
182: 0.25~Hz, there are artefacts at $0.15$ and $0.35$~Hz in the solar
183: wind. We note that in the magnetosheath, where the level of magnetic
184: fluctuations is much higher than in the solar wind, this problem is
185: almost unnoticeable.} operates in the frequency domain
186: $(0.35-12.5)$~Hz. We use FGM data to resolve the frequencies below
187: $0.35$~Hz. Results from Cluster have been compared with data from
188: the Helios~2 satellite within the range $(0.001-1)f_{ci}$. These
189: data are selected within a high-speed stream $V\simeq600$~km/s when
190: the satellite orbited at 0.96 AU \citep{Bruno2003}.
191:
192: %%% FIG 1
193:
194:
195: In order to analyze the magnetic field fluctuations in the solar
196: wind we use the Morlet wavelet transform, defined as
197: \begin{equation}
198: \mathcal{W}_i(\tau,t) = \sum_{j=0}^{N-1} B_i(t_j) \psi [(t_j- t)/\tau]
199: \end{equation}
200: which represents the transform of the $i$-th component of the
201: magnetic field $B_{i}(t_j)$, a data time series with equal time
202: spacing $\delta t$ and $j=0,...,N-1$. Here $\tau$ is a time scale
203: (the corresponding frequency is $f=1/\tau$), while $\psi (u)=
204: 2^{1/2}\pi ^{-1/4} \cos (\omega_0 u) \exp [-u^2/2]$ is the Morlet
205: wavelet, where $\omega_0=6$. The square of a wavelet coefficient
206: gives a ``quantum of energy" of magnetic fluctuations on a scale
207: $\tau$ at a time $t$. Thus we define the power spectral density of
208: $B_i$ as
209: \begin{equation}\label{eq:PSD}
210: \mathcal{S}_i[\textrm{nT}^2/\textrm{Hz}]=\frac{2\delta
211: t}{N}\sum_{j=0}^{N-1}|\mathcal{W}_i(\tau,t_j)|^2.
212: \end{equation}
213:
214: Fig.~\ref{fig:spec} shows the total power spectrum density of the
215: magnetic fluctuations $\mathcal{S}= \sum_{i=x,y,z} \mathcal{S}_i$
216: (solid lines) and the spectrum of magnetic field modulus
217: fluctuations $\delta |{\bf B}|$ (dashed-dotted lines), which
218: corresponds to the spectrum of parallel fluctuations of magnetic
219: field $\mathcal{S}_{\|}$. In agreement with previous observations,
220: the solar wind spectra follow well defined power laws with a break
221: around $f_b\simeq 0.3$~Hz, in the vicinity of $f_{ci}\simeq 0.1$~Hz.
222: Taking into account the Taylor hypothesis, $f_b$ corresponds to a
223: spatial scale $V/f_b \simeq 1800$~km $\simeq 15 \lambda_i$ and to
224: the normalized angular wavenumber $k \lambda _i \simeq 0.4$ ($k
225: \rho_i \simeq 0.3$). Below the break, the Helios spectrum
226: $\mathcal{S} \sim f ^{-1.65}$ is fitted roughly by Kolmogorov's law.
227: The Cluster FGM spectrum covers a $(0.02-0.5)$~Hz frequency range,
228: the lower resolved frequency here is determined by the length of the
229: considered time period, 57 minutes, and by a cone of influence of
230: the Morlet wavelet transform \citep{Torrence1998}. Even if the
231: frequency range is not that large, this spectrum can be described by
232: a power law $\mathcal{S} \sim f ^{-1.75}$. Above the break, the
233: spectrum follows a $\mathcal{S} \sim f ^{-2.6}$ law: the spectral
234: index 2.6 lies in the usual observed range, between 2 and 4
235: \citep{smith06}. As regards the spectrum $\mathcal{S}_{\|}$, it is
236: remarkable, that after the spectral break, the compressible magnetic
237: fluctuations become much more important than in the low frequency
238: part of the spectrum. This can be quantified by the ratio $\langle
239: \mathcal{S}_{\|}\rangle/\langle\mathcal{S}\rangle$, that is
240: $0.02-0.05$ in the law frequency part and $0.26$ in the high
241: frequency part of the spectrum. This means that nearly 30\% of the
242: magnetic fluctuations above the spectral break are compressible.
243:
244:
245:
246: Let us now consider statistical properties of turbulent fluctuations
247: in the high-frequency part of the spectrum. Fig.~\ref{fig:pdfs}
248: shows the probability distribution functions (PDFs) of the
249: normalized variables $b_x(\tau)=\mathcal{W}_x(\tau,t)/\sigma_x$,
250: with the standard deviation $\sigma_x=\left\langle
251: \mathcal{W}_x(\tau,t)^2 \right\rangle^{1/2}$, at three different
252: time scales $\tau$; the dashed lines indicate the corresponding
253: Gaussian fits. Apparently, the smaller the scale, the more the PDFs
254: deviate from Gaussian. The PDFs of the other components of the
255: magnetic field present a similar behaviour. The dependence of
256: normalized PDFs on scale is a signature of intermittency
257: \cite{frisch,noi}.
258:
259: %%% FIG 2
260:
261: Intermittency may be quantified by the flatness. Fig.~\ref{fig:flat}
262: shows the spectrum of flatness $F(f)= \left\langle \mathcal{W}_x^4
263: \right\rangle/\left\langle \mathcal{W}_x^2 \right\rangle ^2$ of
264: $B_x$ fluctuations. We can see that in the low-frequency part $F$
265: increases from the large-scale Gaussian value $F=3$. This means that
266: the usual Alfv\'enic cascade \cite{noi} proceeds towards smaller
267: scales building up phase correlations among fluctuations. In the
268: vicinity of $f_{ci}$, Helios indicates an increase of the flatness
269: and Cluster/FGM shows a slight decrease. This disagreement can be
270: due to (i)~aliasing at the highest frequencies of Helios;
271: (ii)~artifacts on FGM at frequencies around the satellite spin
272: frequency 0.25 Hz; or to some other physical reasons which are
273: outside of the scope of the present paper. Here we will focus our
274: discussion on the higher frequencies, resolved by Cluster/STAFF-SC,
275: where the flatness again displays a power-law dependence on $f$.
276: This scaling behavior is inherent to non-linear dynamics and refers
277: to a new kind of "inertial region".
278:
279: %%% FIG 3
280:
281: In non-magnetized fluid, a rapid exponential increase of the
282: flatness is observed in the near-dissipation range since only the
283: strongest fluctuations survive while the others are destroyed by
284: viscosity \citep{near-diss}. After such increase the flatness
285: saturates. In our case, the saturation is not observed and there is
286: a power law increase of flatness. At the same time strong coherent
287: structures which give an important increase of intermittency
288: represent only $(2-6)\%$ of the fluctuations at each scale (see the
289: insert of Fig. \ref{fig:flat}). This estimation has been made using
290: a method proposed by Farge~(1992), named \emph{local intermittency
291: measure} analysis. It consists in separating the intermittent events
292: which participate to the non-Gaussian tails of the PDF's from the
293: Gaussian fluctuations. At each scale $\tau$ we chose a threshold
294: $\zeta$ for the energy of magnetic fluctuations and we select the
295: wavelet coefficients which verify $|\mathcal{W}(\tau,t)|^2 < \zeta$.
296: With the new set of coefficients we recalculate the flatness; if it
297: is larger than 3 we decrease the threshold until the flatness
298: reaches the Gaussian value. At the end of the procedure, the
299: fraction of the non-Gaussian coefficients gives an estimate of the
300: percentage of the intermittent structures at each scale. Note, that
301: this analysis only gives the upper limit of the structures because
302: one intermittent event is described normally by more than one
303: wavelet coefficient.
304:
305:
306:
307:
308: \section{Role of compressible fluctuations}
309: The results described in the previous section suggest that above the
310: spectral break frequency $f_{b}$ there is a non-linear compressible
311: cascade and not a dissipation range. In this section we show that
312: the level of the plasma compressibility in this range can be at the
313: origin of a \emph{non-universality} of statistics of the magnetic
314: fluctuations. To show this we consider two $7$ min time periods on
315: April 5, 2001, with different ion plasma $\beta_i$.
316:
317: %%% FIG 4
318:
319: Fig.~\ref{fig:compres}(a) shows the total power spectrum density of
320: the magnetic fluctuations within the high frequency range
321: $\mathcal{S}$ (solid line), and the spectrum of parallel
322: fluctuations of magnetic field $\mathcal{S}_{\|}$ (dashed-dotted
323: line) for a time period which starts at 23:03:02~UT, when
324: $\beta_i\simeq 0.5$. Fig.~\ref{fig:compres}(b) shows $\mathcal{S}$
325: and $\mathcal{S}_{\|}$ for a period starting at 22:35:02~UT, when
326: $\beta_i\simeq 1.5$. The straight lines indicate power law fits:
327: $\mathcal{S}\sim f^{-2.33}$ for the first time period and
328: $\mathcal{S}\sim f^{-2.50}$ for the second.
329: Fig.~\ref{fig:compres}(c) shows the ratio
330: $\mathcal{S}_{\|}/\mathcal{S}$ for the two considered time periods.
331: One can see that during the first time period (with $\beta_i \simeq
332: 0.5$) the energy of compressible fluctuations is less than 20\% of
333: the total energy of the fluctuations (see the solid line), while for
334: the second, it is around 30\% (see the dashed line).
335: Fig.~\ref{fig:compres}(d) shows the flatness $F(f)$ calculated
336: within the two time periods (here again the solid line corresponds
337: to the period with smaller $\beta_i$ and dashed line, to the period
338: with higher $\beta_i$).
339:
340: These observations indicate that for higher $\beta_i$, the level of
341: compressibility increases, leading to a steepening of the magnetic
342: power law and an increase of intermittency. To confirm these
343: results more intervals for wider plasma parameters should be
344: analyzed.
345:
346:
347:
348: \section{Nature of the small scale cascade}
349:
350: How can we explain the small scale energy cascade in the solar wind
351: turbulence, sensitive to the compressibility of the fluctuations?
352: There are several interpretations. Following
353: \citep{Schekochihin2007}, at spatial scales between the ion and
354: electron characteristic scales, there is a superposition of
355: (undamped) Kinetic Alfv\'en Wave (KAW) and entropy cascades. KAW
356: waves are expected to have low frequencies $f \ll f_{ci}$ and nearly
357: perpendicular wave vectors with respect to the mean field,
358: $k_{\perp} \gg k_{\|}$.
359:
360:
361: Another possible explanation of the small scale cascade is a
362: superposition of non-linear fluctuations (not waves), which exchange
363: energy on times smaller than the usual eddy-turnover (or non-linear)
364: time. At spatial scales at the order of $\lambda_i$ and at
365: frequencies of the order of $f_{ci}$ one should take into account
366: the Hall effect in Faraday's equation
367: \begin{equation}
368: \frac{\partial \bB}{\partial t} = \rot \left[\bV \times \bB -
369: \frac{m_i}{e \rho}\bJ \times \bB + \eta \bJ \right]
370: \label{induzione}
371: \end{equation}
372: ${\bf J}=c\nabla\times\bB/4\pi$ being the current density and $\eta$
373: a scalar dissipation coefficient. This equation contains three
374: different physical processes, with three associated characteristic
375: times. By introducing for each length-scale $\ell$ the mean value
376: of density $\rho_\ell$, magnetic intensity $B_\ell$ and velocity
377: $u_\ell$, we can define an eddy--turnover time $T_{NL} \sim
378: \ell/u_\ell$, a characteristic time associated to the Hall effect
379: $T_H \sim \rho_{\ell} \ell^2/ B_\ell$ and a dissipative time $T_D
380: \sim \ell^2/\eta$. At large scales the main effect is played by the
381: term $\rot (\bV \times \bB)$, which describes the low-frequency
382: Alfv\'enic cascade \citep{noi} realized in a time $T_{NL}$. We
383: conjecture that approaching the scale of the spectral break, $T_H$
384: appears to be of the order of $T_{NL}$ and beyond the break, the
385: nonlinear energy transfer is realized in a time $\sim T_H < T_{NL},
386: T_{D}$.
387:
388: In a compressible fluid the energy balance equation must be
389: expressed in terms of energy densities (i.e., energy per unit
390: volume) and not in terms of specific energy (i.e., per unit mass),
391: in order to take into account density fluctuations (see
392: \citep{Fleck1996} for example). The mean volume rate of energy
393: transfer on Hall times in a compressible fluid is therefore
394: \begin{equation}\label{eq:e-V}
395: \varepsilon_V \sim \frac{\rho_{\ell} u_{\ell}^2}{T_H}.
396: \end{equation}
397:
398: Assuming equipartition between kinetic and magnetic energies
399: $\rho_{\ell} u^2_{\ell} \sim B^2_{\ell}$, the energy--transfer rate
400: results to be proportional to
401: \begin{equation}\label{eq:e-V-1}
402: \varepsilon_V \sim \frac{B_{\ell}^2}{T_H} \sim
403: \frac{B_{\ell}^3}{\ell^2\rho_{\ell}}.
404: \end{equation}
405:
406: Following von Weizs\"{a}cker (1951), the ratio of the mass density
407: $\rho$ at two successive levels of the hierarchy is related to the
408: corresponding scale size $\ell$ by the following equation
409: \begin{equation}\label{eq:rho-turb}
410: \frac{\rho_{\ell_{\nu}}}{\rho_{\ell_{\nu+1}}} = (\frac{\ell_{\nu}}
411: {\ell_{\nu+1}})^{-3\alpha}
412: \end{equation}
413: where $|\alpha|$ is a measure of the degree of compression at each
414: level $\nu$ (larger $\nu$ meaning larger length-scale), and ranges
415: from $\alpha=0$ for no compression up to $|\alpha|=1$ for isotropic
416: compression ($3|\alpha|$ is a number of dimensions in which the
417: compression takes place). So, using this density scaling,
418: $\rho_{\ell} \sim \ell ^{-3\alpha}$, and assuming a constant
419: spectrum energy transfer rate we have
420: \begin{equation}\label{eq:b-turb}
421: B_{\ell} \sim \rho^{1/3} \ell^{2/3} \sim \ell^{2/3 - \alpha}.
422: \end{equation}
423: Therefore, the spectral energy function is
424: \begin{equation}\label{eq:E-turb}
425: E(k) \sim \frac{B_{\ell}^2}{k} \sim k^{-7/3 + 2\alpha}.
426: \end{equation}
427:
428: In the incompressible plasma limit ($\alpha=0$), this phenomenology
429: predicts a $\sim k^{-7/3}$ spectrum. Such a spectrum has been
430: observed both in direct numerical simulations of an incompressible
431: Electron MHD turbulent system \citep{Biskamp1996} and in the EMHD
432: limit of the incompressible Hall MHD shell model \citep{Buchelin06}.
433:
434: In the case of isotropic compression toward smaller scales
435: ($\alpha=1$) which can take place in the interstellar medium, the
436: spectrum is $E(k)\sim k^{-1/3}$. If the isotropic compression is
437: going on toward larger scales ($\alpha =-1$) the spectrum will be
438: $E(k)\sim k^{-11/3}$. In the case, analyzed here, the power law
439: spectrum of magnetic fluctuations follows the $\sim k^{-2.6}$--law,
440: that corresponds to $\alpha \simeq -0.14$, i.e., the scaling
441: relation for the density is $\rho_{\ell} \sim \ell ^{0.4}$.
442:
443: This simple phenomenological model allows to explain the variations
444: of the spectral index of the high frequency part of the solar wind
445: spectrum from $-4$ to $-2$ by different degree of plasma compression
446: with $\alpha$ between $-5/6$ and $1/6$.
447:
448: However, the described model have to be improved in order to take
449: into account (i) the space anisotropy ($k_{\perp}\neq k_{\|}$) that
450: appears in a plasma with a mean field and (ii) possible different
451: scaling for velocity and magnetic field which can be found in the
452: compressible Hall MHD \citep{sergio07}. Besides, the interpretation
453: of the small scale energy cascade in the Hall MHD frame is supported
454: by the observations of a clear correlation between the spectral
455: break frequency $f_b$ and the Doppler-shifted wavevector of the ion
456: inertia length $f_{\lambda_i}$ (Leamon et al., 2000).
457:
458:
459:
460: Note that the solar wind spectral break frequency $f_b$ is usually
461: observed to be lower than $f_{\lambda_i}$ \citep{Leamon98}. It can
462: be explained by the fact that the Hall effect starts acting at
463: scales larger than $\lambda_i$. In our case, for example,
464: $f_b/f_{\lambda_i} \simeq 0.1$ and the scale of the spectral break
465: is $\sim 10 \lambda_i$.
466:
467: Introducing the Hall Reynolds
468: number as the ratio of the non-linear term over the Hall term
469: \begin{equation}
470: R_H \sim \frac{e}{c m_i} \frac{\rho_{\ell} u_{\ell} \ell}{B_{\ell}}
471: \end{equation}
472: one can see that $R_H \simeq 1$ (i.e. $T_{NL} \simeq T_{H}$) at a
473: scale $\ell$, that is not necessarily equal to $\lambda_i$,
474: $$\frac{\ell}{\lambda_{i\ell}} =4\pi \frac{V_{A\ell}}{u_{\ell}} $$
475: here $V_{A\ell}$ and $\lambda_{i\ell}$ are the mean values of the
476: Alfv\'en speed and the ion inertial length at scale $\ell$
477: respectively.
478:
479:
480: \section{Conclusion}
481:
482: In this paper we investigate small scales turbulent fluctuations in
483: the solar wind (i.e. at frequencies above the spectral break at the
484: vicinity of $f_{ci}$). Taking into account the Taylor hypothesis
485: ($\ell=V/f$) the frequency domain above the spectral break covers
486: $\sim (40- 2000)$~km space scales that corresponds to $(0.3-
487: 15)\lambda _i$. In this range we found evidences for strong
488: departure from Gaussian statistics and the presence of intermittency
489: while the spectrum presents a well-defined power law. Both the
490: presence of a power-law spectrum and the absence of global
491: self-similarity, seems to be quite in contrast with the role of
492: ``dissipative range".
493:
494: In usual fluid turbulence, the dissipative range \citep{frisch}
495: starts with a rough exponential cutoff; in the near dissipation
496: range the intermittency increases as far as the Gaussian
497: fluctuations dissipate faster than the coherent structures
498: \citep{near-diss}; then the fluctuations become self-similar, the
499: singularities being smoothed by dissipation. In the solar wind
500: turbulence we observe a completely different picture. After the
501: spectral break in the vicinity of $f_{ci}$ the flatness increases as
502: a power law indicating that non-linear interactions are at work to
503: build up a new inertial range.
504:
505: This small scale cascade is much more compressible than the lower
506: frequency Alfv\'enic cascade. This sudden change in nature of the
507: turbulent fluctuations can happen due to a partial dissipation of
508: magnetic fluctuations at the spectral break: the left-hand
509: Alfv\'enic fluctuations with $k_{\|}\gg k_{\perp}$ are damped by the
510: ion-cyclotron damping \citep{Ghosh96}. Above the break, however, a new
511: \emph{magnetosonic cascade} takes place up to the electron
512: characteristic scales. This energy cascade is seems to be dominated
513: by the fluctuations with $k_{\perp}\gg k_{\|}$
514: \citep{fouad06,Mangeney2006}.
515:
516: We found, as well, that the plasma compressibility controls the
517: statistics of the magnetic field fluctuations. Preliminary results
518: show that the increase of the level of the plasma compressibility
519: leads to the spectrum steepening and increase of the intermittency.
520:
521: To explain this small scale compressible cascade we introduce here
522: for the first time a simple phenomenological model based on the
523: compressible Hall MHD: we find that the magnetic energy spectrum
524: follows a $E(k) \sim k^{-7/3 + 2\alpha}$ law, depending on the
525: degree of plasma compressibility $\alpha$. While far from a complete
526: model of small scale turbulence, this simple model can explain
527: observed variations of the spectral index within the high frequency
528: part of the solar wind turbulent spectrum \citep{Leamon98,smith06}
529: by different degree of plasma compressibility.
530:
531:
532:
533: \acknowledgments
534: We acknowledge discussions with A.~Mangeney, R. Grappin,
535: R.~Bruno and S.~Servidio. We thank N.~Cornilleau, E.~Lucek and
536: I.~Dandouras for providing Cluster data. OA acknowledges the support
537: of CNES and European network TOSTISP.
538:
539:
540: \begin{thebibliography}{}
541:
542: \bibitem[Bale et al. 2005]{Bale2005}
543: Bale, S.~D., Kellogg, P.~J., Mozer, F.~S., Horbury, T.~S., and Reme,
544: H. 2005, Phys. Rev. Lett., 94, 215002
545:
546: \bibitem[Balogh et al. 2001]{Balogh2001}
547: Balogh, A. \textit{et al.} 2001, Ann. Geophys., 19, 1270
548:
549: \bibitem[Biskamp et al. 1996]{Biskamp1996}
550: Biskamp, D., Schwarz, E., and Drake, J.~F. 1996, Phys. Rev. Lett.
551: 76, 1264
552:
553: \bibitem[Boyd and Sanderson 2003]{HMHD}
554: Boyd, T.~J.~M., and Sanderson, J.~J. 2003, The Physics of Plasmas,
555: Cambridge Univ. Press
556:
557: \bibitem[Bruno and Carbone 2005]{noi}
558: Bruno, R., and Carbone, V. 2005, Living Rev. Solar Phys., 2, 4
559:
560: \bibitem[Bruno et al. 2003]{Bruno2003}
561: Bruno, R., Carbone, V., Sorriso-Valvo, L. and Bavassano, B. 2003, J.
562: Geophys. Res., 108, 8
563:
564: \bibitem[Chevillard et al. 2005]{near-diss}
565: Chevillard, L., Castaing, B., and L\'ev\^eque, E. 2005, Eur. Phys.
566: J. B, 45, 561
567:
568: \bibitem[Cornilleau-Wehrlin et al. 2003]{Cornilleau2003}
569: Cornilleau-Wehrlin, N. \textit{et al.} 2003, Ann. Geophys., 21, 437
570:
571: \bibitem[Farge 1992]{Farge1992}
572: Farge, M. 1992, Ann. Rev. Fluid Mech. 24, 395
573:
574: \bibitem[Fleck 1996]{Fleck1996} Fleck, R.~C., Jr.\ 1996, \apj,
575: 458, 739
576:
577: \bibitem[Frisch 1995]{frisch}
578: Frisch, U. 1995, Cambridge Univ. Press
579:
580: \bibitem[Galtier 2006]{Galtier06}
581: Galtier, S. 2006, J. Plasma Phys., 72, 721
582:
583: \bibitem[Galtier and Buchlin 2007]{Buchelin06}
584: Galtier, S., and Buchlin, E. 2007, \apj, 656, 560
585:
586: \bibitem[Ghosh et al. 1996]{Ghosh96}
587: Ghosh, S. \textit{et al.} 1996, J. Geophys. Res., 101, 2493
588:
589: \bibitem[Leamon et al. 1998]{Leamon98}
590: Leamon, R.~J., Smith, C.~W., Ness, N.~F., Matthaeus, W.~H., and
591: Wong, H.~K. 1998, J. Geophys. Res., 103, 4775
592:
593: \bibitem[Leamon et al. 1999]{Leamon99}
594: Leamon, R.~J. \textit{et al.} 1999, J. Geophys. Res., 104, 22331
595:
596: \bibitem[Leamon et al. 2000]{Leamon2000}
597: Leamon, R.~J. \textit{et al.} 2000, \apj, 537, 1054
598:
599: \bibitem[Li et al. 2001]{Li01}
600: Li, H., Gary, S.~P., and Stawicki, O. 2001, Geophys. Res. Lett., 28,
601: 1347
602:
603: \bibitem[Mangeney et al. 2006]{Mangeney2006}
604: Mangeney, A. \textit{et al.} 2006, Ann. Geophys., 24, 3507
605:
606: \bibitem[R\'eme et al. 2001]{CIS2001}
607: R\'eme, H. \textit{et al.} 2001, Ann. Geophys., 19, 1303
608:
609:
610: \bibitem[Sahraoui et al. 2006]{fouad06}
611: Sahraoui, F. \textit{et al.} 2006, Phys. Rev. Lett., 96, 075 002
612:
613:
614: \bibitem[Servidio et al. 2007]{sergio07}
615: Servidio, S. \textit{et al.} 2007, Planet. and Space Sci., in press
616:
617: \bibitem[Schekochihin et al. 2007]{Schekochihin2007}
618: {Schekochihin}, A.~A., {Cowley}, S.~C., {Dorland}, W. 2007, Plasma
619: Phys. Control. Fusion 49 A195-A209
620:
621:
622: \bibitem[Smith et al. 2006]{smith06}
623: Smith, C.~W., Hamilton, K., Vasquez, B.~J., and Leamon, R.~J. 2006,
624: \apj, 645, L85
625:
626: \bibitem[Stawicki et al. 2001]{stawicki}
627: Stawicki, O., Gary, S.~P., and Li, H. 2001, J. Geophys. Res., 106, 8273
628:
629: \bibitem[Torrence and Compo 1998]{Torrence1998}
630: Torrence, C. and Compo, G.~P., 1998, A Practical Guide to Wavelet
631: Analysis, Bulletin of the American Meteorological Society, vol.~79,
632: Issue 1, pp.61-78
633:
634: \bibitem[Tu and Marsch 1995]{sw2}
635: Tu, C.~-Y., and Marsch, E. 1995, Kluwer Academic Pub.
636:
637: \bibitem[Weizs\"{a}cker 1951]{Weizsacker1951}
638: von Weizs\"{a}cker, C.~F. 1951, \apj, 114, 2, 165
639:
640: \end{thebibliography}
641: \clearpage
642:
643: \begin{figure}
644: \epsscale{.80} \plotone{f1.eps} \caption{The total magnetic field
645: power spectral density $\mathcal{S}$ and the spectrum of
646: compressible magnetic fluctuations $\mathcal{S}_{\|}$ (dashed-dotted
647: line) measured by Helios 2 (up to $0.08$ Hz) and by Cluster (up to
648: $12.5$~Hz). Straight lines refer to power law fits. Vertical dotted
649: lines indicate the ion cyclotron frequency $f_{ci}$ and the
650: Doppler-shifted ion inertial length $f_{\lambda_i}$.}
651: \label{fig:spec}
652: \end{figure}
653:
654:
655: \begin{figure}
656: \epsscale{.80} \plotone{f2.eps}\caption{The probability distribution
657: functions (PDFs) of the normalized variable $b_x(\tau)$ at three
658: time scales: $\tau = 2$ s (upper panel), $\tau = 0.7$ s (middle
659: panel), and $\tau = 0.2$ s (bottom panel). For each panel a
660: Gaussian function is plotted for reference as a dashed curve.}
661: \label{fig:pdfs}
662: \end{figure}
663:
664:
665: \begin{figure}
666: \epsscale{.80} \plotone{f3.eps}\caption{The flatness of the magnetic
667: field component $B_x$ calculated using Cluster data (solid line) and
668: Helios 2 data (dashed-dotted line); the horizontal solid line
669: indicates the Gaussian value $F=3$. The insert gives the percentage
670: of intermittent events which participate to the flatness in the high
671: frequency part of the spectrum.} \label{fig:flat}
672: \end{figure}
673:
674: \begin{figure}
675: \includegraphics[angle=90,scale=.80]{f4.eps} \caption{Properties of magnetic
676: fluctuations above the spectral break for different plasma
677: $\beta_i$; (a) $\mathcal{S}$ and $\mathcal{S}_{\|}$ for a 7
678: minutes time period with $\beta_i\simeq 0.5$; (b) the same as (a)
679: but for a 7 minutes time period with $\beta_i\simeq 1.5$; (c) level
680: of compressible fluctuations for period (a), solid line, and (b),
681: dashed line; (d) flatness of the fluctuations of $B_x$ component
682: calculated for the period (a), solid line, and for the period (b),
683: dashed line.} \label{fig:compres}
684: \end{figure}
685:
686:
687:
688: \end{document}
689: