Turbulence, nonlinear dynamics, and sources of

Turbulence, nonlinear dynamics, and sources of
intermittency and variability in the solar wind
Intermittency & turbulence
“Intermittency is the nonuniform distribution of eddy formations in a stream. The modulus or
the square of the vortex field, the energy dissipation velocity or related quantities quadratic in
the gradients of Velocity and Temperature (of the concentration of passive admixture) may serve
as indicators. “ (E A Novikov, J Appl Math & Nech, 35, 266 (1971)
Intermittency in simple form
•
•
•
•
Duffing oscillator
Lorenz attractor
Rikitake dynamo
many others
Spatial fluctuations of
dissipation are very large – gradients
Are not uniformly distributed;
the cascade produces intermittency
Some types of intermittency and potential effects on solar
prediction
(1) Large scale/low frequency intermittency
- variability of sources
- Inverse cascade (space)  1/f noise (time)
- Effects of dynamics on the “slow manifold”
 Dynamo reversals, rare events (big flares?)
(2) Inertial range intermittency
- “scaling” range
- reflects loss of self similarity at smaller scales
- KRSH
 This is a lot of what you see and measure
(1) Dissipation rage intermittency
- vortex or current sheets or other dissipation structures
- usually breaks self similarity because there are characteristic physical scales
 Controls local reconnection rates and local dissipation/heating;
small scale “events”
Langmuir cells
•
•
•
•
•
Turbulence
Waves
Structure
Gradients
Mode coupling
Intermittent turbulence in hydro
• Dynamics at cloud tops:
temperature gradients,
driven by droplets
(J. P. Mellado, Max Plank Meteor.
• ocean surface-air
Interface (J P Mellado)
• Vorticity
In interstellar
Turbulence
(Porter,
Woodward,
Pouquet)
• PDFs
intermittency corresponds to “extreme
events,” especially at small scales
 fat tails
• Higher order moments and nonGaussianity (esp.
increments or gradients)
For Gaussian, odd moments zero,
Even moments < 2 > determined by <x2>;
For intermittency, <x2n> > Gaussian value
• Kurtosis and filling fraction F
κ= < 4>/ <x2>2
HEURISTIC: κ ∼ 1/F
energy
containing
inertial
dissipation
energy
input
cascade
heating
Energy spectrum E(k)
“standard” turbulence spectrum
Log(wavenumber)
• Dissipation: conversion of (collective) fluid degrees of freedom into
motions into kinetic degrees of freedom
• Heating: increase in random kinetic energy
• Entropy increase: irreversible heating
How nonlinearity and cascade produces intermittency
concentration of gradients
• Amplification of higher order moments
–
Suppose that q & w are Gaussian, and
 

∼
then pdf( ) is exponential-like with κ � 6-9.


• Amplification greater at smaller scale ( e. g.,
• Role of stagnation points (coherency!)

∼

– No flow or propagation to randomize the concentrations
• Formation is IDEAL
 ,wavenumber )
(e.g., Frisch et al. 1983; Wan et al, PoP 2009)
• Dissipation is more intense in presence of gradients  relation between
intermittency and dissipation
Coherent structures are
generated by ideal
effects!
dissipative
ideal
Contours of current density:
Already have non-Gaussian
coherent structures 
…before finite resolution
errors set in 
 TIME
Same initial condition 
Turbulent fluctuations have structure
and dissipation is not uniform
ε : dissipation rate; ∆v: velocity increment
Kolmogorov ’41
∆v ∼ (εr) 1/3
 <∆vp > = const. ε p/3 r p/3
Kolmogorov ’62
∆v ∼ (εr r)
1/3
But this is NOT observed!
εr = r-3 ∫ d3x’ ε(x’)
Kolmogorov refined similarity hypothesis
 <∆vp > = const. <εr p/3 > r p/3
= const. ε
p/3
r
p/3 + ξ(p)
(Oubukhov ’62)
multifractal theory
comes from this!
Intermittency in hydrodynamics • Anselmet et al, JFM 1984
Need to be sure
Pdf is resolved well
enough to compute
higher order moments!
<∆urn> ∼ rζ(n)
Pdfs of longitudinal velocity increments
have fat tails; fatter for smaller scales
ζ(n)
Scaling of exponents at increasing order:
reveals departures from self similarity and
multifractal scalings (beta, log-normal,
She-Levesque, etc
n
SW/MHD intermittency
• More dynamical variables
• Analogous effects
Intermittency in MHD & Solar wind
• Multifractal scalings
• PDFs of increments
(Politano et al, 1998; Muller and Biskamp 2000)
(Burlaga, 1991; Tu & Masrch 1994, Horbury et al 1997)
Muller & Biskamp, 2000
Sorriso et al, 1999
Cellularization, turbulent relaxation and structure in plasma/MHD:
large scale evolution produces local relaxation  suppression of
nonlinearity  nonGaussian statistics  boundaries of relaxed regions
correspond to small scale intermittent structures
• Local relaxation can give rise to
• Force free states
• Alfvenic states
• Beltrami states
AND
• characteristic small scale intermittent
structures , e.g. current sheets

v
- Simulations show RAPID relaxation & production of local correlations.
- Spatial “patches” of correlations bounded by discontinuities.
θ
b
Characteristic distributions appear
in less than one nonlinear time!
Directional alignment: pdf
Run with Hc
f(cos(vb))
0
v-b correlations: large (black >0; white < 0 )
(here, 2D MHD)
• Analysis of patches of Alfvenic correlations
•
•
•
•
Distributions of cos(θ ) [angle between velocity and magnetic field]
Global statistics & statistics of linear subsamples (∼1-2 correlation scales)
SW and 3D MHD SIM (512^3)
Linear SIM samples
10 hr SW samples
Global Alfvenicity σc ≈ 0.3
- For a specified sample size, can get highly variable
Alfvenicity (see Roberts et al. 1987a,b)
- Same effect in SW and in SIMs!
PVI Coherent Structure Detection: designed to work the
same way in analysis of solar wind and simulation data
 =
•
|∆  , |
<|∆ , |>/
∆ (x;s) = (x + s) – B(x)
•
•
•

Greco et al, GRL 2008; ApJ 2009
PVI links classical discontinuities and intermittency
& compares well between SW and simulations
Sim
Greco et al, ApJ 2009;
Servidio et al JGR,
2011
PVI > 3 events are statistically
inconsistent
with Gaussian statistics at the
90+ % level
distribution of PVI
SW
500 correlation scale
PVI time series
Waiting time distribution between “events”
PVI vs. classical
discontinuity
methods
PVI events in SW
And in MHD turbulence
simulations
• Use PVI to find reconnection sites
• In SIMs & in SW (caveats)
From Servidio et al, JGR,
116, A09102 (2011)
Trajectory thru SIM 
”time series” of PVI
Condition is
PVI > threshold (1,2,3…
↓ Get a
Table of efficiencies
At PVI>7
- only ID ~40% of reconnection sites
- But >95% of events are reconnection
sites
Same approach in SW, but
compare t Gosling/Phan identified
exhaust events:
PVI>7 event in SW very likely to
be at/near a reconnection event!
Osman et al, PRL 112, 215002
(2014)
Evidence that coherent structure are sites of enhanced heating:
Solar wind proton temperature distribution conditioned on
 =
|∆  , |
<|∆ , |>/
Similar (weaker)
effects in:
electron temp
&
electron
heat flux
ALSO: neighborhoods of larger PVI events are hotter
Wind s/c
Osman et al, 2011
Osman et al, 2012
20
Implications for energetic
particle transport
Transport boundaries are observed:
“dropouts” of Solar energetic particles
core” ofH-FE
SEP
with
ions vs
arrival dropouts
time
For 9 Jan 1999 SEP event
From Mazur et al, ApJ (2000)
plasma intermittency
• At kinetic scales
• Still more variables, but analogous effects
Localized kinetic effects in 2.5D Eulerian Vlasov simulation
(undriven initial value problem; strongly turbulent )
•
Magnetic field, current density, X
points
•
Kinetic effects near a “PVI event”
Greco et al, PRE, 86, 066405 (2012)
Servidio et al, PRL 108, 045001 (2012)
Out-of-plane
current
•
Anisotropy Tmax/Tmin in small area
•
•
•
•
a) nonMaxwellianity
b) proton T anisotropy
c) proton heat flux
D) kurtosis of f(v)
There is a strong association of kinetic effects with current structures!
Dissipation is concentrated in sheet-like structures
in kinetic plasma
Wan, Matthaeus, Karimabadi, Roytershteyn, Shay, Wu ,
Daughton, Loring, Chapman, 2012
Strength of electric current density in shear-driven kinetic
plasma (PIC) simulation (see Karimabadi et al, PoP 2013)
Thinnest sheets seen are comparable to electron inertial length. Sheets are clustered
At about the ion inertial length  heirarchy of coherent, dissipative structures at kinetic scales
Scale dependent kurtosis:
MHD, kinetic sims, SW comparison
Wu et al, ApJ Letters
763:L302012 (2013)
Very low frequency/very large scale intermittency
•
1/f noise:
–
–
–
–
–
•
Gives “unstable” statistics – bursts and level-changes
Long time tails on time correlations
Generic mechanisms for its production (Montroll & Schlesinger, 1980)
Often connected with inverse cascade, quasi-invariants,
highly nonlocal interactions (opposite of Kolmogorov’s assumption!)
Dynamo generates 1/f noise (experiments: Ponty et al, 2004
• connected to statistics of reversals (Dmitruk et al, 2014)
– 1/k  1/f inferred from LOS photospheric magnetic field
– 1/f signature in lower corona
– 1/f signatures observed in density and magnetic field in solar wind
at 1 AU (M+G, 1986; Ruzmaiken, 1988; Matthaeus et al, 2007; Bemporad et al, 2008)
An example from 3D MHD with strong mean magnetic field
(Dmitruk & WHM, 2007)
-
nearly in condensed state
-
energy shifts at times scales
of 100s to 1000s Tnl
-
characteristic Tnl ~ 1
-
Where do these timescales
come from ?
Numerical experiments on
MHD Turbulence with mean field: onset of 1/f
noise due to “quasi-invariant”
    Increasing B0
behavior of
a Fourier mode
In time, from
simulation
Eulerian frequency spectra
Eulerian frequency
spectrum:
transform of
one point
two time
Correlation fn.
B0=8
Dmitruk &
Matthaeus,
2009
1/f noise in SW (1AU ISEE-3, OMNI datasets)
Matthaeus & Goldstein, PRL 1986
1/f: 1AU, MDI and UVCS – high/low latitude comparisons
Ulysses
MDI
UCVS
Matthaeus et al, ApJ 2007
Bemporad et al, ApJ 2008
1/f noise and
reversals in spherical
MHD dynamo
Incompressible MHD
spherical Galerkin model
low order truncation
 Run for 1000s of Tnl
 See ramdon reversals
of the dipole moment
 1/f noise with rotation
and or magnetic helicity
Dmitruk et al, PRE in
press 2014
With rotation/helicity  Waiting times for reversals scale like geophysical data!
possible 1/f
In time domain
cascade
energy
containing
Slow &
incoherent
nonuniform
dissipation
intermittency
corrections!
Faster
more Coherent
more nonGaussian
heating
Energy spectrum E(k)
More detailed cascade picture: central role of
Log(wavenumber)
• Cascade: progressively enhances nonGaussian character
• Generation of
and patchy correlations
• Coherent structures are sites of
• for inverse cascade/quasi-invariant case, 1/f noise low frequency
irregularity in time, and build up of long wavelegnth fluctuations
Toy model to generate
intermittency
- May be useful in transport studies as an
improvement over random phase data
- We already saw that structure is generated
by ideal processes…so…
Synthetic realizations with intermittency
• Minimal Lagrangian Map (Rosales & Meneveau,
2006)
• Add magnetic field; map using velocity (Subedi et al,
2014)
Choose spectrum
Iterate low pass filtering
get filtered fields
Push filtered vector fields
v & b with filtered v-field
at this level
Re-map onto grid by
averaging; proceed
to next level
After several (M=7) iterations
Pdfs of
longitudinal
magnetic
increments
vs lag.
Perpendicular
current
density
in a plane
Comparison of scale
dependent kurtosis:
SW, synthetic and MHD
simulation
summary
Intermittency is a factor in solar prediction and space weather:
• Large scale/low frequency intermittency (1/f noise) controls
unsteady fluctuations in global parameters including extreme events
• Inertial range intermittency generates structures that channel, trap and transpsort
SEPs and change connectivity of field lines
• Small scale (kinetic) intermittency implements heating and dissipation and controls
reconnection rates
Coherent magnetic structures emerge in many
theoretical models
Current and
Magnetic field
in 2D MHD
simulation
Parker problem: RMHD
Rappazzo & Velli 2010
3D isotropic
MHD current
Mininni,
NJP 2008
3D Hall MHD compressible,
strong B0, current
Dmitruk 2006
2.5D kinetic hybrid
Parashar et al, 2010