Transcript Solar flares

Solar Flares
Lyndsay Fletcher
University of Glasgow
Introduction
Part 1: Observations
energy build-up
impulsive phase
atmospheric response
Part 2: Theory
nature of reconnection
particle acceleration
PPARC/Mallorca
September 2006
1
RHESSI and TRACE flare observations
Bremsstrahlung emission:
12-25keV – Red
25-50keV – Blue
Thermal emission at ~ 1.5MK and ~25MK, green
PPARC/Mallorca
September 2006
2
Flare Questions
• What is the source of the flare energy?
• How is energy stored and how/why is it released?
• How is energy converted into heat, particles, radiation, bulk
kinetic energy?
• How does the solar atmosphere respond?
PPARC/Mallorca
September 2006
3
Introduction
Part 1: Observations
energy build-up
impulsive phase
atmospheric response
Part 2: Theory
nature of reconnection
particle acceleration
PPARC/Mallorca
September 2006
4
Source of Flare Energy
Energy is imparted to the coronal magnetic field at or below photosphere photosphere is a high beta plasma so gas pressure forces dominate.
magnetic field
evolution from
SOHO/MDI
PPARC/Mallorca
September 2006
5
Energy storage in the corona
MHD version of Ampère’s Law
j = B/m
Twisting the field produces ‘free energy’ in the form of current.
MHD Force balance equation
 

Dv
  p  j  B  ρg  
Dt
Assume ~ steady state, with negligible gravitational forces and pressure
gradients (low beta corona). Then
 
Force-free condition, i. e.
jB  0
field and current are aligned
So


 
  B  B  0, meaning


  B  αB
a constant along field lines
PPARC/Mallorca
September 2006
6
When does energy build-up occur?
Does the magnetic field emerge already bearing free energy?
From vector magnetic field measurements, Leka et al. (1996) determined
the curl of the magnetic field in an emerging active region
They also measured the photospheric flows during and after emergence.
Result:
AR twists are too large to be
generated by the photospheric flows.
So the field emerges from the
convection zone already carrying
current.
PPARC/Mallorca
September 2006
7
Location of free energy - observations
Shear is concentrated
close to magnetic
neutral line
PPARC/Mallorca
September 2006
8
Can flares be predicted?
Folk Knowledge:
A complex, rapidly-evolving, large active region has highest probability of
producing a flare, within a few days of its emergence.
More accurate predictions?
Prediction based on past X-ray activity (Bayesian statistics)
Moderately successful (Wheatland 2004)
Statistics of magnetic field parameters and their variations
Rather unsuccessful (Leka and Barnes 2006)
Neural Net ‘learning’ of appearance of ARs about to flare
Underway
PPARC/Mallorca
September 2006
9
Preflare Activity
What are the signs that a flare is going to happen?
Most Reliable - The rise/darkening/expansion of an AR Ha filament
minutes to hours before flare (Svestka 1976, Martin 1980)
PPARC/Mallorca
September 2006
10
Filaments are dense, cool gas suspended in the corona.
The slow filament rise probably indicates the onset of an MHD instability.
polar crown filament
active region filament
movie
movie
PPARC/Mallorca
September 2006
11
Preflare Activity
Are there any other signs that a flare is going to happen?
Other pre-flare phenomena include:
•
•
•
•
•
Small UV/EUV ‘twinkles’ (Moore & Sterling, Warren & Warshall)
Small GOES events and preheating
“Sigmoid” magnetic configuration (Hudson & Sterling)
Early hard X-ray coronal sources (Lin et al.)
Moving blueshifted Ha events (Des Jardins & Canfield 2003)
But none of these is unique to flares
PPARC/Mallorca
September 2006
12
Energy storage and flare trigger - overview
• Magnetic energy for the flare is stored as field-aligned currents.
• Most of the energy is concentrated close to a neutral line, probably
low in the atmosphere (~10,000km).
• The flare is related to the onset of an MHD instability, probably
followed by fast magnetic reconnection.
• The conditions leading to a flare cannot easily be identified – flare
prediction is still not possible.
PPARC/Mallorca
September 2006
13
Introduction
Part 1: Observations
energy build-up
impulsive phase
atmospheric response
Part 2: Theory
nature of reconnection
particle acceleration
PPARC/Mallorca
September 2006
14
Flare X-ray Time Profiles
Impulsive phase - energy release
• Hard X-rays (10s of keV)
• Duration ~ 5 minutes to 1 hour
• Bursty time profile (tacc ~ seconds)
Gradual phase - response
• thermal emission (~0.1-1 keV)
• rise time ~ minutes
GOES flare classification scheme
Flux in the 1-8 Å band
more than 10-6 W m-2 at Earth
(e.g. C4.2 flare => 4.2x10-6 Wm-2)
M: more than 10-5 W m-2 at Earth
X: more than 10-4 W m-2 at Earth
C:
PPARC/Mallorca
September 2006
15
Impulsive Phase X-ray Spectrum
hot thermal emission
Non-thermal electron
bremsstrahlung: I(e) ~ e-g
Gamma-ray
lines
PPARC/Mallorca
September 2006
16
X-Rays
X-rays observed by e.g. RHESSI are electron-proton bremsstrahlung from
energetic electrons (> 15keV)
• Thermal bremsstrahlung: Eelectron ~ E target and spectrum F(e)~ e-e/kt
- Primarily coronal.
• Non-thermal bremsstrahlung: Eelectron >> E target and spectrum F(e)~eg
- Primarily chromospheric ‘thick target’ (electron slows as it radiates)
- Very inefficient: ~ 10-5 of the electron energy radiated as X-rays.
PPARC/Mallorca
September 2006
17
RHESSI X-ray imaging
• 2” imaging at ~ 20-50 keV, ~ 4s resolution (if sufficient counts)
• Spectroscopy with 1 keV energy resolution
Orange = 25-50keV
Blue = TRACE white light
20” =15,000km
PPARC/Mallorca
September 2006
RHESSI imaging
18
X-Ray spectrum
• In impulsive phase, HXR spectrum can be fitted by a hot (20MK) or
superhot (~60MK) thermal component plus a power-law F(E)=FoEg.
• Parameters of fit are correlated and vary during flare.
• Microflares also show thermal + power-law spectrum.
electron spectral index
10
E (keV)
100
Typical spectral fit, Holman et al (2003)
PPARC/Mallorca
September 2006
Variation of fit parameters during a
flare, Grigis & Benz (2005)
19
HXR spectral inversion
Photon spectrum I(e) is related to source-averaged electron spectrum F E 
If Q is known, the integral can be inverted (i.e. differentiated) to recover
source-averaged spectrum (Kontar et al. ’03, ’06)
No photospheric albedo correction
With albedo correction
Collisional thick target interpretation 
Beam number ~ 1036-37 e- s-1 above 20keV
PPARC/Mallorca
September 2006
Combining with observed HXR footpoint
areas < 10” x 10” 
Beam flux > 1018-19 e- cm-2 s-1 above2020keV
Other Impulsive Phase Emission
During the impulsive phase bursts of g-rays hard X-rays, UV/EUV, Ha and
(sometimes) optical emission show where fast particles hit the low atmosphere
1600 A broadband emission
HXR contours on 195A emission
Optical/UV/EUV emission from heat
deposition /ionisation / collisional /
radiative excitation
White light luminosity increase confirms
total energy input deduced from HXRs
PPARC/Mallorca
September 2006
White light footpoints
21
What is significant about footpoint locations?
Radiation from non-thermal particles appears at the photospheric
intersection of separatrix surfaces (eg Demoulin, Aulanier).
Motion of footpoints across magnetic field can be used to deduce a
coronal magnetic reconnection rate.
Separatrix intersections
Time evolution of HXR
footpoint positions
Metcalf et al. (2003)
PPARC/Mallorca
September 2006
22
Separatrices (2-D and 3-D)
Separatrices are (curves)/surfaces separating domains of different
magnetic connectivity in (2D)/3D
Radiation produced at predicted separatrix locations shows the
importance of reconnection processes happening at domain boundaries
(Priest and Schrijver 1999)
(Priest and Schrijver 1999)
PPARC/Mallorca
September 2006
23
Flare Impulsive Phase Energy Spectrum
hot thermal emission
Non-thermal electron
bremsstrahlung: I(e) ~ e-g
Gamma-ray
lines
PPARC/Mallorca
September 2006
24
Gamma-rays
• Continuum g-rays by bremsstrahlung (~ 10MeV)
• Nuclear de-excitation lines caused by
proton bombardment;
- ‘prompt’ radiation provides a diagnostic
of protons above 30MeV
Production of nuclear de-excitation lines
• The positron annihilation line at 511keV
• The neutron capture line at 2.23 MeV - n(p,g)D
- this is a delayed line, as neutrons must slow down before reacting.
 formed low in the atmosphere, and after other emissions.
PPARC/Mallorca
September 2006
25
Location of g-ray sources
Similar time profiles for HXRs and de-excitation lines imply related
acceleration.
However, radiation produced by ions is in a different location from electron
bremsstrahlung
Neutrons produced by energetic ions
(10s of MeV/nucleon).
The capture line is predicted to form
within 500km of neutron production
site.
Hurford et al. 2006
PPARC/Mallorca
September 2006
But observed to be systematically
offset from HXRs (electrons) by ~
10,000km (15”).
26
Flare Energy Budget
Estimates made by Emslie et al. (2004, 2005) confirm that a significant
fraction of total flare energy appears in fast particles
above 20 40keV
PPARC/Mallorca
September 2006
27
Coronal radio emission
Metric and decimetric Type III bursts are plasma radiation produced by
electron beams (mode-conversion of Langmuir waves).
Upward and downward-going beams observed
Occur at peak time of HXR emission. Detailed radio/HXR burst-to-burst
time-correlations improve at higher starting frequency (Benz et al ‘05)
Non-linear coherent emission  electron number estimates difficult.
Microwave gyro-synchrotron
emission from fast electrons in
coronal loops is also significant.
Electron numbers can be made
to agree with numbers from HXR.
Also info on angular distribution
PPARC/Mallorca
September 2006
28
Coronal Hard X-ray Sources
 First observed by SMM/HXIS, then at better resolution by Yohkoh/HXT.
Now in many RHESSI flares (inc. occulted)
 Flares seen without HXR footpoints – only coronal HXR loops. Emission
up to 50keV implies ne ~ 1011cm-3 (Veronig & Brown 2004)
Occulted sources present up to at least 50 keV (Bone et al. 2006)
20-30 keV
30-50 keV
50-100 keV
Coronal energy deposition between 1030 and 1031 erg
PPARC/Mallorca
September 2006
29
In situ electron measurements
Some hard X-ray events are also observed by particle detectors in space
(e.g. 14 Apr 2002, Krucker et al, 2003)
In situ electron fluxes do not
agree with thick-target fluxes
deduced from RHESSI
14-April-2002, 22:25 UT
10
WIND/3DP, d=2.6
<F(E)>, RHESSI, d=2.5
F 0(E), RHESSI, d=4.5
1
Electron Flux
0.1
0.01
1E-3
1E-4
1E-5
10
100
Energy, keV
Impulsive phase – main points
• During the impulsive phase, most of the flare energy is released.
• Energetic particles receive up to 50% of the total energy (rest goes
to CME and heating).
• Most of the impulsive phase energy is focused into a few small
sources.
• These sources are closely related to magnetic separatrices.
• Up to 1037 electrons accelerated per second (i.e. all the electrons in
(10,000km)3 at 1010 cm-3.
• Similar number of ions accelerated, but in a different place or
following a different path.
PPARC/Mallorca
September 2006
31
Introduction
Part 1: Observations
energy build-up
impulsive phase
atmospheric response
Part 2: Theory
nature of reconnection
particle acceleration
PPARC/Mallorca
September 2006
32
Atmospheric response – soft X-ray and EUV
Flare leads to brightening coronal loops, producing high fluxes of soft Xray emission (0.1-1 keV).
loops then cool
through EUV
temperatures
PPARC/Mallorca
September 2006
33
Atmospheric response - chromospheric upflows
Energy input in the form of a beam heats the chromosphere rapidly and
– if heating is strong enough - it expands upwards.
Chromospheric ‘evaporation’ has been
observed spectroscopically in Ha (e.g.
Antonucci et al 1984)
Also in UV/EUV using SoHO/CDS postflare observations (e.g. Czaykowska et
al 1999)
Blueshifts on outer part of arcade
are evaporation
Redshifts on inner part of arcade
are material cooling and draining.
PPARC/Mallorca
September 2006
34
Latest numerical simulations disagree with evaporation scenario
Coronal density increases, but temperature does not: increasing
ionisation, radiative losses and gas expansion absorb the energy
T
ne
Allred et
al 2006.
PPARC/Mallorca
September 2006
35
Introduction
Part 1: Observations
energy build-up
impulsive phase
atmospheric response
Part 2: Theory
nature of reconnection
particle acceleration
PPARC/Mallorca
September 2006
36
Liberation of stored energy
Magnetic reconnection allows the coronal field to reconfigure, liberating
magnetic energy
Reconnection is the process whereby two field lines, being frozen in
and carried along by the fluid, break and rejoin in a different way.
A
B
So a particle that was on fieldline A can end up on fieldline B
Reconnection results from the local breakdown of flux conservation.
PPARC/Mallorca
September 2006
37
Breakdown of Flux Conservation
The induction equation – describes advection and dissipation of field


 
B
2
   (v  B)  η B
t
Define associated timescales :
ta = L/v and
td = L2/h
where L is the typical length scale for variation in the magnetic field.
The ratio of td to ta is called the Magnetic Reynolds number, RM
vL
τD
RM 

τA
η
Normally in the corona, dissipation is much slower than advection (low
collisional resistivity h, large v and L); so the field is frozen.
PPARC/Mallorca
September 2006
38
Magnetic Field Dissipation
td in the corona is about 106 years. How do we speed up dissipation?
td = L2/h so must decrease the length scale L, or increase resistivity.
As field is advected by flow, it generates steep gradients - current sheets
V
Field lines are advected in to the current sheet, reconnect, and plasma is
advected out at the upstream Alfvén speed - Sweet-Parker reconnection.
Sweet-Parker reconnection is rather inefficient: the rate scales as Rm-1/2
It is too slow (by ~ 5 orders of mag) to explain flare energy release.
PPARC/Mallorca
September 2006
39
Petschek Reconnection
The Petschek model reduces the size of the diffusion region.
Reconnection rate increases as plasma is ‘slingshotted’ out
Outflow at
~ Alfvèn speed
• The reconnection rate is determined by the external conditions
• At slow shocks, energy conversion can occur
• May be fast enough to explain flare release if a high ‘anomalous resistivity’
is invoked.
PPARC/Mallorca
September 2006
40
Non-MHD regimes of reconnection
Most solar reconnection theory is done in the MHD framework.

  
E  v  B  hj
MHD Ohm’s law
This only includes a resistivity term due to classical (Ohmic) dissipation.
Generalised Ohm’s law
 1  m m n    j 
 
  

i
e
E  v  B  hj 

    mi  me  j  B  me pi  mi p e 

e  e  t  n 

e.g. the Hall term arises since protons and electrons follow different orbits
in a magnetic field
• Decoupling of e, p leads to whistler wave generation
• Dissipation occurs by wave-particle interaction
PPARC/Mallorca
September 2006
41
2-D versus 3-D reconnection
Simple models are 2D. The corona is certainly 3D
(Priest and Schrijver 1999)
(Priest and Schrijver 1999)
PPARC/Mallorca
September 2006
42
Introduction
Part 1: Observations
energy build-up
impulsive phase
atmospheric response
Part 2: Theory
nature of reconnection
particle acceleration
PPARC/Mallorca
September 2006
43
Mechanisms of Particle Acceleration
Accelerating coronal charged particles requires electric fields.
There are three mechanisms most often discussed in the context of
acceleration of solar flare particles
•DC field acceleration - E large-scale and organised
•Stochastic resonant acceleration – E small-scale and random
•Diffusive Shock Acceleration – Both large and small-scale fields
The last of these three is generally thought of as a ‘secondary’ accelerator
of particles, in the presence of a shock associated with a coronal mass
ejection. We shall concentrate on the DC and stochastic models.
PPARC/Mallorca
September 2006
44
Field-aligned DC electric fields
• For example, a local increase in resistivity in a current loop leads to
large potential drop (since huge inductance of circuit prevents rapid
change in current)
• Electrons accelerated if this DC field is greater than the Dreicer field, Ed
The Dreicer field is the value of the DC field such that the force exerted
on the electrons exceeds drag force from e-e Coulomb collisions
• Ed typically 10-4 V/cm
• Electrons with speeds greater than a critical speed =vTe (Ed/E)1/2 are
freely accelerated  ‘runaway’ electrons
electron beam
FE
PPARC/Mallorca
September 2006
B
FDRAG
45
DC electric fields in a current sheet
Observations suggest that super-Dreicer fields occur in reconnection regions.
Inferred values of reconnection electric field are ~ 1 V/cm
In a model with symmetry in the plane perpendicular to the loop, this field is
• parallel to solar surface and
• normal to loops in post-flare arcade
Problem: E  B almost everywhere, so
mostly particles EB drift.
E
However, near X-line, there may be an
unmagnetised region (B(x,y)  0)
- or a component of B parallel to E
L~109cm
total potential drop = 1010 V!
PPARC/Mallorca
September 2006
So some efficient particle acceleration
can occur – how much?
46
Stochastic Resonant Acceleration
Particles gyrating in a magnetic field pick up energy from plasma waves
if their gyration frequency is resonant with (multiples of) the wave frequency
Frequency matching condition:
wk|| v|| - lW/g = 0
w,W = wave and cyclotron frequencies
k|| = parallel wave number
v|| = parallel particle speed
l = integer
However, as soon as a particle picks up some energy from a monochromatic
wave, its gyration frequency changes and resonance is lost.
But particles resonating with a wave spectrum can ‘hop’ stochastically
from one resonance to another as their energy increases or decreases
(e.g. Miller & Vinas 1993, Miller 1998)
PPARC/Mallorca
September 2006
47
Energy vs. time for a single proton in isotropic, high frequency turbulence
Energy lost or gained
in each interaction, but
overall, the energy of
the particle increases
Miller & Viñas (1993)
Protons and ions
(mostly) resonate with
high-frequency Alfvenic
waves
Electrons resonate with
electron cyclotron waves
PPARC/Mallorca
September 2006
48
Closing Thoughts
Observations of solar flares in the X-ray and EUV, from the last 15 years
or so, have demonstrated a great variety and complexity
To a great extent, they still support the overall picture of reconnection in the
corona resulting in the acceleration of a vast number of particles, which lead
to heating and mass-motion in the lower atmosphere.
However, the study of solar flares involves some formidable theoretical
problems, such as:
• What is the microphysics of coronal reconnection?
• How do we extend our understanding of 2-D reconnection to 3-D?
• How do we tie together the MHD, plasma and kinetic aspects of theory?
We must also not ignore the physics which has been painfully learned in other
fields, such as lab devices and magnetospheric reconnection
PPARC/Mallorca
September 2006
49
RHESSI preflare sources
RHESSI often observes X-rays
at ~ 10-20 keV, several minutes
before the impulsive phase.
11:04
PPARC/Mallorca
September 2006
11:06
11:08
11:10
11:12
11:14
50
Coronal Sigmoids
So-called for their ‘S’ shape (in the Northern Hemisphere)
Visible in soft X-rays (e.g. by Yohkoh/SXT and now GOES SXI)
pre-flare sigmoid observed by
Yohkoh Soft X-ray Telescope
PPARC/Mallorca
September 2006
post-flare configuration – cusp
shows that an eruption has
happened.
51
RHESSI Imaging
RHESSI uses ‘collimating optics’ - slits and slats
Only X-rays from certain directions can be detected
Rotation axis
Germanium detector
PPARC/Mallorca
September 2006
52
How RHESSI makes an image
Slats rotate past
source
Detector
Illumination
Modulation
patterns
Modulated brightness patterns from all 9 grid pairs are combined
to calculate the most likely source intensity distribution
Technique is similar to radio interferometry
PPARC/Mallorca
September 2006
53
Coronal ‘Free Energy’
field ‘twist’  ‘free energy’ in the form of current: j = B/m


  B  α(x, y) B
a = 0:
‘potential’ field. There are no currents (no free energy)
a = const:
linear force-free field. j= aB
a  const:
non-linear FFF
MDI/SXT
PPARC/Mallorca
September 2006
potential
nonlinear FF
58
Coronal HXR sources
gradual variations
below 10 keV
night
> 20keV emission faint,
fast time variations
night
Nov 18, 2003: GOES M4, Krucker
PPARC/Mallorca
September 2006
Upwards motion ~5-40
km/s
Coronal energy deposition
between 1030 and 1031 erg
(depending on model)
60