Dissipation-scale turbulence in the solar wind

G. G. Howes, S. C. Cowley, W. Dorland, G. W. Hammett, E. Quataert, and A. A.

Schekochihin

Citation: 932, 3 (2007); doi: 10.1063/1.2778938

View online: http://dx.doi.org/10.1063/1.2778938
View Table of Contents: http://aip.scitation.org/toc/apc/932/1
Published by the American Institute of Physics
 Dissipation-scale turbulence in the solar wind
G. G. Howes*, S. C. Cowley^ W. Dorland**, G W. Hammett^ E. Quataert*
and A. A. Schekochihin^
'Department of Astronomy, University of California, Berlceley, CA, USA.
^Department ofPtiysics and Astronomy, UCLA, Los Angeles, CA, USA.
"Department of Physics and Center for Scientific Computing and Mathematical Modeling,
University of Maryland, College Park, MD, USA.
^Princeton Plasma Physics Laboratory, Princeton, NJ, USA.
^King 's College, Cambridge, UK and Department of Physics, Imperial College, London, UK.

Abstract. We present a cascade model for turbulence in weakly coUisional plasmas that follows
the nonlinear cascade of energy from the large scales of driving in the MHD regime to the small
scales of the kinetic AlfVen wave regime where the turbulence is dissipated by kinetic processes.
Steady-state solutions of the model for the slow solar wind yield three conclusions: (1) beyond the
observed break in the magnetic energy spectrum, one expects an exponential cut-off; (2) the widely
held interpretation that this dissipation range obeys power-law behavior is an artifact of instrumental
sensitivity limitations; and, (3) over the range of parameters relevant to the solar wind, the observed
variation of dissipation range spectral indices from —2 to —4 is naturally explained by the varying
effectiveness of Landau damping, from an undamped prediction of —7/3 to a strongly damped index
around —4.
Keywords: Plasma turbulence—solar wind—kinetic damping—gyrokinetics
PACS: 52.35.Ra, 52.30.Gz, 96.25.Qr, 96.50.Tf

INTRODUCTION
One of the principal measurements in the study of solar wind turbulence is the magnetic
field fluctuation frequency spectrum derived from in situ satellite measurements. At
1 AU, the one-dimensional energy spectmm in spacecraft-frame frequency typically
shows, for low frequencies, a power law spectrum with slope of -5/3, suggestive of a
Kolmogorov-like inertial range [1,2]; a spectral break is typically observed at around 0.4
Hz, with a steeper power law at higher frequencies, often denoted the dissipation range in
the literature, with a spectral index that varies from -2 to -4 [3,4]. The general consensus
is that the -5/3 portion of the spectmm is the inertial range of an MHD turbulent cascade,
but the dynamics responsible for the break and steeper portion of the spectrum is not well
understood. Various explanations for the location of the break in the spectrum have been
proposed: that it is coincident with the proton cyclotron frequency in the plasma [5, 3, 6],
or that the fluctuation length scale has reached either the proton Larmor radius [7, 8] or
the proton inertial length [9, 10]. The steepening of the spectmm at higher wavenumbers
has been attributed to proton cyclotron damping [1, 5, 3, 6], Landau damping of kinetic
AUven waves [7, 8, 9], orthe dispersive nature of whistler waves [11].
To unravel the underlying physical mechanisms at work in the solar wind requires an
understanding of turbulence in weakly collisional, magnetized plasmas. Early theories
of MHD turbulence proposed an isotropic cascade of turbulent energy [12, 13], but nu-
CP932, Turbulence and Nonlinear Processes in Astrophysical Plasmas—
6* Annual International Astrophysical Conference, edited by D. Shaikh and G. P. Zank
© 2007 American Institute of Physics 978-0-7354-0443-4/07/$23.00
merical simulations [14] demonstrated an inherent anisotropy in the presence of a mean
magnetic field. An evolving anisotropic theory [15, 14, 16, 17, 18] has emeiged which
rests upon two central hypotheses: the Kolmogorov hypothesis of locality in wavenum-
ber space [19], and the conjecture that in strong turbulence the linear wave periods main-
tain a critical balance with the nonlinear turnover timescales. The anisotropic nature of
the turbulence means the frequency for nonlinear energy transfer is dominated by the
perpendicular wavenumber, k^v^, where _L denotes the component perpendicular to the
mean magnetic field. Assuming balanced turbulence with equal Elsasser energy fluxes in
either direction along the mean field, the one-dimensional magnetic energy spectmm in
—5/3
the MHD regime scales asEB{kj^)ock^ ' and critical balance implies a scale-dependent
2/3

anisotropy with hi oc k^ [18]. In the regime of electron MHD (EMHD) [20], one ob-
tains EB{ki_) oc k'l^l^ and k\\ oc k^l^ [21, 22].
Are observations of turbulence in the solar wind consistent with these theoretical pre-
dictions? The eneigy in turbulent fluctuations is observed to be anisotropic [23] with
k^>k\\ in the slow solar wind at scales of ^^p,--^ 10^^ [24], where p, is the proton Lar-
mor radius; this appears consistent with the prediction of a scale dependent anisotropy
leading to nearly perpendicular wavevectors ^^ > ^|| at smaU scales. The imbalance be-
tween anti-sunward and sunward Elsasser spectra can reach nearly two orders of magni-
tude in the fast wind, while the slow wind has a much smaller imbalance, from a factor
of a few to approximate equality [25, 26]. Thus, we beheve the aforementioned theory
of MHD turbulence to be relevant to the dynamics in the slow solar wind.
Although the large scales at which the turbulence is driven may be adequately de-
scribed by MHD, the turbulent fluctuations at the small-scale end of the inertial range
often have parallel wavelengths smaller than the ion mean free path; therefore, a kinetic
description of this weakly collisional plasma is required to capture the turbulent dynam-
ics. The slow, fast, and entropy modes are damped in a warm, colhsionless plasma [27];
the Alfven wave cascade, however, is undamped untfl it reaches the ion Larmor radius,
k^Pi r^ 1 [28, 29]. For a sufficiently large inertial range, wavevectors at this scale be-
come nearly perpendicular with kx_ > k\\; thus, frequencies remain low compared to the
ion cyclotron frequency (a < Q.i, the nonlinear cascade to yet smaller scales is composed
of kinetic AUven waves, and Landau damping by the ions and electrons can effectively
dissipate the turbulence. The dynamics in this regime optimally described by a low-
frequency limit of kinetic theory called gyrokinetics [30, 29]. Here we present a model
aimed at following the nonlinear cascade of magnetic energy from fluid to kinetic scales
while accounting for the kinetic dissipation of the turbulence.

ANALYTICAL MODEL
Consider a homogeneous magnetized plasma with a mean magnetic field of magnitude
BQ that is stirred isotropically at an outer scale wavenumber ko with velocity VQ. We write
the magnetic field fluctuations in velocity units, b^ = 5 5 ^ (^±) / v'47r«,/w,. The frequency
of nonlinear eneigy transfer for Alfvenic fluctuations at a given perpendicular wavenum-
ber is estimated to be co„i '-^ k^y^ = k^bk. Assuming the locality of nonlinear interactions
in wavenumber space and a constant energy cascade rate e, the one-dimensional mag-
netic energy spectrum in the regime k^pi <C 1 is given by

(1)
K\

where Ci„ is a dimensionless constant of order unity. The frequency of nonlinear energy
transfer is
(0„i=C2„s'/^kf, (2)
where C2m is another order unity constant; the paraUel wavenumber can be determined
by applying the critical balance conjecture, setting the linear Alfven wave frequency
equal to the nonlinear frequency co = co„i.
In the kinetic AllVen wave regime k^pi > 1, the dynamics are governed by
the equations of Electron Reduced MHD [29], with characteristic fluctuations
Vk = ±bkk^pi/ y^Pi + 2/(1 + Te/Ti). Applying the same procedure for this regime
yields the one-dimensional magnetic energy spectrum

(3)

and the nonlinear frequency

2/3
,4/3
(4)
[/3,+2/(i + 7;/7;)]i/3

A continuity equation for the magnetic energy per unit mass at each wavenumber b^
can be written as [31]

-'^+Sik,)-2l^^co,ik,)bl (5)
dt dlnk± (o{k±)

where the three terms on the right-hand side are the energy flux through wavenumber
space, a source term, and a damping term. The energy cascade rate is modeled by

^_3, c-,\k^p,r v
£{ki_)=k^bl (6)
1" li, + 2/{l + Te/T,)

and the nonlinear frequency by

1/2
elm , C\ [k^p,f
0)nl{kl_) =k^bk
Q™ Q i A + 2 / ( i + 7;/7 ;)J (7)

In the damping term, j/o) is determined from the hnear gyrokinetic dispersion relation
[30]. The order unity constants are taken to be C\m = Cik = 2.5 and C2m = Q i = 2.2
based on numerical simulation, as in Quataert and Gruzinov [32].
 1 1 ' '" iiii 1
1 '
Panel ( b )
0.1 \Xock-=/3 \
0.01
: NX 1

"1
0.001 r 1
0.0001 \«k-'/3
r "1
> 10-=
1 ^
\VVck-3
^ io-« ; 1
to ?,
10-' r 1
...1 . 1 .
10-
100 0.01 0.1 1 10 100

FIGURE 1. One-dimensional magnetic energy spectra for three gyrokinetic models: (1) j3, = 0.5,
Ti/Te = 3, (2) ft = 3, Ti/Te = 0.6, (3) ft = 0.03, Ti/Te = 0.175. All models use ytop, = 3 x 10"^ Panel
(a) shows that all three spectra demonstrate a dissipative roll-oif with a variation of spectral indices in
the range k±pi > 1. Panel (b) adds a constant magnetometer sensitivity limit to each spectrum, yielding
dissipation range spectra that more closely resemble power laws with a range of slopes from —7/3 to —4

RESULTS
The model given by equation (5) is solved numerically to obtain a steady state magnetic
energy spectrum for a given set of the parameters ion plasma beta /3,, ion to electron
temperature ratio Ti/Te, and isotropic driving scale kopi. Panel (a) of Figure 1 presents
the solutions for three cases chosen to sample the observed parameter range in the
solar wind [3, 33, 34]: (1) /3, = 0.5, T,/Te = 3; (2) /3, = 3, T,/Te = 0.6; and (3) /3, =
0.03, Ti/Te = 0.175. All models use kopi = 3 x 10^^. In the absence of dissipation,
analytical theory predicts spectral indices of —5/3 in the MHD regime and —7/3 in the
kinetic AlfVen wave regime; with the damping rate artificially set to zero, this model
indeed recovers these results (not shown). Damping at k^pi > 1 is sufficient to cause
each spectmm in panel (a) to fall off more steeply than the undamped prediction of
—7/3. The steady-state spectra obtained here clearly demonstrate the exponential roll-
off characteristic of dissipation [35].
Observations of the magnetic fluctuation spectra at higher wavenumbers than the
spectral break are widely interpreted to behave like a power law rather than an exponen-
tial decay. We suggest here that the power-law appearance of the spectrum in this range
is an effect of limited magnetometer sensitivity; this sensitivity hmit can be clearly seen
in Figure 6 of Leamon et al. [3] at the high wavenumber end of the spectmm. The noise
floor of a fluxgate magnetometer at frequencies / > 1 Hz is constant in units of n T / y ^
[36], so we mock up the instrumental noise by specifying a constant background value
of the one-dimensional eneigy spectmm. We choose the noise floor to be approximately
two to three orders of magnitude below the spectmm value at the break. Panel (b) of
Figure 1 adds a constant sensitivity level at two (spectrum 1) or three (spectra 2 and
3) orders of magnitude below the spectmm value at k^pi = 1. The behavior of each
spectmm in panel (b) in the range k^pi > 1 more closely resembles a power law than
the exponential roll-off in the noiseless spectra; the steady-state solutions are well-fit by
power laws with spectral indices —7/3, —3, and —4. In summary, the instmmental sen-
sitivity limit is cmcial in interpreting measured magnetic fluctuation spectra, and may
produce spectra that appear to obey a power-law scaling even though the underlying
spectmm is actually rolling off exponentially.
The spectral index in the dissipation range is observed to vary from —2 to —4 [3, 4].
Figure 1 shows that, over the range of the plasma parameters /3, and Ti/Te measured in
the solar wind, this variation can naturally be explained by the varying effectiveness of
the damping of kinetic Alfven waves via the Landau resonance. If Landau damping is
neghgible, the spectral index is expected to give a value of —7/3, close to the observed
upper limit; over the range of parameters relevant to the solar wind, this cascade model
gives a lower limit to the spectral index of about —4, for example spectmm 3, consistent
with observations. Hence, the varying effectiveness of Landau damping is sufficient to
explain the observed variation of spectral indices, with the break occurring at the ion
Larmor radius.

CONCLUSION
The physical mechanisms responsible for the spectral break and steeper dissipation
range of the magnetic energy spectmm observed in the solar wind have not been con-
clusively identified. This paper presents a turbulent cascade model constmcted to follow
the magnetic fluctuation energy from the laige scales in the MHD regime down to the
small scales in the kinetic Alfven wave regime, accounting for dissipationby kinetic pro-
cesses. Due to the inherent anisotropy of MHD turbulence, the tuibulence remains low
frequency co <C O; and is optimally described by gyrokinetics. This picture of balanced,
low-frequency turbulence is relevant to the slow solar wind.
The nonlinear cascade model given by (5) using the gyrokinetic damping rates is
solved numerically to find steady-state solutions as presented in Figure 1. The cascade
model predicts that, for wavenumbers above the break in the magnetic fluctuation en-
eigy spectrum, the spectmm undergoes a slow exponential cut-off. We aigue that the
widespread interpretation that this dissipation range shows power-law behavior is an ar-
tifact of limited magnetometer sensitivity. Over the range of parameters /3, and Ti/Te
measured in the solar wind, the varying strength of Landau damping naturally repro-
duces the observed variation of dissipation range spectral indices from —7/3 to —4,
with the spectral break occurring at the scale of the ion Larmor radius.
This model assumes that linear damping rates are relevant for turbulent fluctuations
that are nonlinearly cascaded to smaller scales on the timescale of one wave period.
Nonlinear gyrokinetic simulations of the turbulent cascade in the transition to the kinetic
AUven wave regime are necessary to judge the validity of this assumption. This cascade
model can be used as a tool to connect nonlinear numerical simulations to observations
of turbulence in the solar wind. Further work to examine the importance of the proton
cyclotron resonance in dissipation of solar wind turbulence is underway [34].
 ACKNOWLEDGMENTS
Thanks to S. Bale for helping to apply this work to the solar wind. This work was
supported by the DOE Center for Multiscale Plasma Dynamics, Fusion Science Center
Cooperative Agreement ER54785 and the hospitality of the Aspen Center for Physics.

REFERENCES
1. p. J. Coleman, Jr., Astrophys J. 153, 371 (1968).
2. M. L. Goldstein, D. A. Roberts, and W. H. Matthaeus, Ann. Rev. Astron. Astrophys 33, 283-326
(1995).
3. R. J. Leamon, C. W. Smith, N. F. Ness, W. H. Matthaeus, and H. K. Wong, J. Geophys Res 103,
4775^787(1998).
4. C. W. Smith, K. Hamilton, B. J. Vasquez, and R. J. Leamon,Astmphys J. Lett 645, L85-L88 (2006).
5. M. L. Goldstein, D. A. Roberts, and C. A. Fitch, J. Geophys Res 99, 11519-11538 (1994).
6. S. R Gary, J. Geophys Res 104, 6759-6762 (1999).
7. R. J. Leamon, W. H. Matthaeus, C. W. Smith, and H. K. Wong, Astrophys J. 507, L181-L184 (1998).
8. R. J. Leamon, C. W. Smith, N. F. Ness, and H. K. Wong, J. Geophys Res 104 22331-22344 (1999).
9. R. J. Leamon, W. H. Matthaeus, C. W Smith, G. R Zank, D. J. Mullan, and S. Oughton, Astrophys J.
537, 1054-1062(2000).
10. C. W Smith, D. J. Mullan, N. F. Ness, R. M. Skoug, and J. Steinberg, J. Geophys Res 106, 18625-
18634(2001).
11. O. Stawicki, S. R Gary, and H. Li, J: Geophys Res 106,8273-8282(2001).
12. R. S. Iroshnikov,^5/ro«. Zh 40, 742- (1963), english Translation: Sov. Astron., 7 566 (1964).
13. R. H. Krakhnan,Phys Fluids S, 1385-(1965).
14. J. V. Shebahn, W. H. Matthaeus, and D. Montgomery, J. PlasmaPhys 29, 525-547 (1983).
15. D. Montgomery,PhysicaScripta2, 83-88 (1982).
16. J. C. BligAon, Astrophys J. 285, 109-123 (1984).
17. S. Sridhar, and R GO\AKK\I, Astrophys J. 433, 612-621 (1994).
18. R Goldreich, and S. SnAhar, Astrophys J. 438, 763-775 (1995).
19. A. N. Kolmogorov, Dokl. Akad. Nauk SSSR 30, 9 - (1941), english Translation: Proc. Roy. Soc.
LondonA,434,9(1991).
20. A. S. Kingsep, K. V. Chukbar, and V. V. Yankov,i?ev. PlasmaPhys 16, 243 (1990).
21. D. Biskmap, E. Schwarz, A. Zeiler, A. Celani, and J. F Drake, Phys Plasmas 6, 751-758 (1999).
22. J. Cho, and A. Lazarian, Astrophys J. Lett 615, L41-L44 (2004), a s t r o - p h / 0 4 065 95.
23. W. H. Matthaeus, M. L. Goldstein, and D. A. Roberts, J. Geophys Res 95, 20673-20683 (1990).
24. S. Dasso, L. J. Milano, W H. Matthaeus, and C. W. Smith, Astrophys J. Lett. 635, L181-L184
(2005).
25. C.-Y. Tu, E. Marsch, and H. Rosenbauer, Geophys Res Lett 17, 283-286 (1990).
26. R. Grappin, A. Mangeney, and E. Marsch, J. Geophys Res 95, 8197-8209 (1990).
27. A. Barnes, Phys Fluids 9, 1483-1495 (1966).
28. E. Quataert, Astrophys J. 500, 978-+(1998), a s t r o - p h / 9 7 1 0 1 2 7 .
29. A. A. Schekochihin, S. C. Cowley, W. Dorland, G. W Hammett, G. G. , Howes, and E. Quataert,
Astrophys J. Supp. (2007), submitted, a r X i v : 0704 . 0 0 4 4 v l .
30. G. G. Howes, S. C. Cowley, W. Dorland, G. W Hammett, E. Quataert, and A. A. Schekochihin,
Astrophys J. 651, 590-614 (2006), a s t r o - p h / 0 5 1 1 8 1 2 .
31. G. K. Batchelor, The Theory of Homogeneous Turbulence, The Theory of Homogeneous Turbulence,
Cambridge: Cambridge University Press, 1953, 1953.
32. E. Quataert, and A. Gmzinov, Astrophys J. 520,248-255 (1999), a s t r o - p h / 9 8 03112.
33. J. A. Newbury, C. T. Russell, J. L. Phillips, and S. R Gary, J. Geophys Res 103, 9553-9566 (1998).
34. G. G. Howes, S. C. Cowley, W. Dorland, G. W Hammett, E. Quataert, and A. A. Schekochihin,
J. Geophys. Res. (2007), in preparation.
35. H. Li, S. R Gary, and O. Stawicki, Geophys Res Lett 28, 1347-1350 (2001).
36. S. D. Bale, private communication (2007).