rhessi imaging spectroscopy a regularization ... - unipa.itdaa_erice07/contributors/massone.pdf ·...
TRANSCRIPT
Electron flux maps of solar flares: a regularization approach to
RHESSI imaging spectroscopy
Anna Maria Massone
CNR-INFM LAMIA, Genova, Italy
In collaboration with :
• Michele Piana, Dipartimento di Informatica, Università di Verona, Italy
• Marco Prato, Dipartimento di Matematica Pura ed Applicata, di Modena e Reggio Emilia, Italy
• Gordon Emslie, Dep. of Physics, Oklahoma State University, US
• Gordon Hurford, Space Sciences Laboratory, University ofCalifornia at Berkeley, US
• Richard Schwartz, NASA Goddard Space Flight Center, Greenbelt, US
• Eduard Kontar, Dep. of Physics & Astronomy, The University, Glasgow, UK
People involvedPeople involved
The RHESSI mission has been launched by NASA on February 5, 2002in order to understand the high-energy processes at the core of the solar flare phenomena :
A solar flare is the most energetic explosion in the solar system.
The energy released during a flare is typically on the order of 1027 erg per
second (~1019 KW). Large flares can emit up to 1032 erg of energy.
Significant electromagnetic emission, particularly in the X-ray range.
RHESSI MissionRHESSI Mission((Reuven Ramaty High Energy Solar Spectroscopic Imager))
Scientific objective: to study the processes of electron and ion acceleration in solar flares through the hard X-ray and gamma-ray radiation that the
accelerated particles produce
RHESSI observationsRHESSI observations
• High spatial resolution X-ray images
� arc-sec-quality images
� image restoration: CLEAN, MEM, forward-fitting
• High energy resolution X- and γ-ray spectra
� 1 keV spectral resolution
� energy range: 1 keV – 17 MeV
High-resolution spectroscopy at each point of the X-ray imageImaging Spectroscopy
e +H
e
X-ray
spectra/images/imaging spectroscopy
RHESSI
Photon space Photon space vsvs Electron spaceElectron spaceX-ray emission: Bremsstrahlung
1. Photon spectra electron spectra (inversion of the Bremsstrahlung eq.)
2. Photon images local photon spectra local electron spectra
Is it possible to build images in the electron space?
• nine bi-grid collimators• nine Ge detectors
� planar array of equally spaced, X-ray-opaque slats separated by transparent slits
� nine different widths for the slit
RHESSI HardwareRHESSI Hardware
• θ is the incident angle between the photons and the collimator axis
• duration of a completeRHESSI rotation: 4s
As the spacecraft rotates, imaging information is encoded as rapid time-variations of the detected flux
RHESSI GeometryRHESSI Geometry
Modulation profilesModulation profiles
point source
different θ
more off-axis
source distribution
bigger source distribution
real profile
smaller intensity
Roll angle and aspect phaseRoll angle and aspect phase
xsun
ysun
Spacecraft axis
α
β
• roll angle (α): defines the grid direction with respect to acoordinate system
For each detector:
{ }sunsun yx ,
• aspect phase (β) : measures the position of a reference point near the source with respect to the spacecraft axis
α
β
tq ↔
( )βαt ,↔
( )βαq ,↔Data stacking
Data stacking IData stacking I
RHESSI data are light curves, i.e. photon-induced counts
recorded while the collimators rotate
For each time point, a roll angle
and a phase are definedco
unts
coun
tsco
unts
roll
angl
e (d
eg)
phase (deg)
For different rotations, differentphases correspond to the same
roll angle
The counts corresponding tothe same roll and phase bin
are stacked in a histogram (32 rollbins, 12 phases)
Data stacking IIData stacking II
Data stacking IIIData stacking III
For each roll bin, the count profile, as a function of the phase bin, is fitted by a
Fourier series
If A and B are the first two Fourier components, the complex number whose real part is A and
imaginary part is B is called visibility
Visibilities IVisibilities I
(u,v) spatial frequency components(x,y) point in the sourceI(x,y;ε) photon spectrum at point (x,y) and energy εV(u,v;q) observed visibility at energy qD(qj,εk) Detector Response Matrix (DRM)
Formal definition:
A RHESSI visibility is a complex observable number that can be derived from RHESSI data and which represents a
measurement of a single Fourier component of the source distribution measured at a specific spatial frequency and
energy- and time-range.
∫∫∑+=
yx kk
vyuxiπkkjj dxdyεeεyxIεqDqvuV ∆),,(),(),,( )(2
where
∫∫∑+=
yx kk
vyuxiπkkjj dxdyεeεyxIεqDqvuV ∆),,(),(),,( )(2
∫∞
=ε
dEEεQEyxFyxNRπ
εyxI ),();,(),(41);,( 2
∫∞
=q
dEEqKEvuWqvuV ),();,();,(
∫∫+=
yx
vyuxiπ dxdyeEyxFyxNEvuW )(2);,(),(:);,(
kk
kkjj EQqDR
EqK εεεπ ∑ ∆= ),(),(41:),( 2
Count Count �� Electron Visibilities IElectron Visibilities I
Electron Visibilities
∫∞
=q
dEEqKEvuWqvuV ),();,();,(
Count Count �� Electron Visibilities IIElectron Visibilities II
Visibility inversion problem: determine the electron visibilities, W(u,v;E), from the
observed count visibilities V(u,v;q)
The relation between the measured count visibilities and the electron visibilities is described by a Volterra integral equation of the first kind
{ }N1=kkkk v,u;σ
2. Fix λ by means of some optimality criterion
The Singular Value Decomposition of the operator A gives the set of triplets:
The regularized solution: ‡”1
2 )(N
kkk
k
kλ uvg
λσσW
=
⋅+
=
VAW =
The Tikhonov Method:
1. Solve the minimum problem min- 22 =+ WλVAW
fidelity smoothness
Solution strategy: Tikhonov regularization method
Visibility inversionVisibility inversion
∫∞
=q
dEEqKEvuWqvuAW ),();,();,)((
The The algorithmalgorithm
2002, February 20 - 11:02:08 – 11:14:20 UT
Time range selected: 11:06:02 – 11:06:34 UT
• For each detector and each (u,v) point:
� construct the count visibility spectrum (count visibility vs count energy)
� apply regularized inversion to obtain an electronvisibility spectrum (electron visibility vs count energy)
• For each detector and each electron energy:
� construct the electron visibilities
Selected flare:
Count energy range selected: 10 – 50 keV
Count energy binning selected: 4 keV
RHESSI data: RHESSI data: countcount visibilitiesvisibilities
Detector 1
(u1(1),v1
(1)) Re(V(u1(1),v1
(1);q)) Im(V(u1(1),v1
(1);q))
(u32(1),v32
(1)) Re(V(u32(1),v32
(1);q)) Im(V(u32(1),v32
(1);q))
Detector 9
(u1(9),v1
(9)) Re(V(u1(9),v1
(9);q)) Im(V(u1(9),v1
(9);q))
(u32(9),v32
(9)) Re(V(u32(9),v32
(9);q)) Im(V(u32(9),v32
(9);q))
Fixed energy channel [q,q+∆q]
VisibilityVisibility--basedbased countcount mapsmaps
10-14 keV 14-18 keV 18-22 keV 22-26 keV 26-30 keV
30-34 keV 34-38 keV 38-42 keV 42-46 keV 46-50 keV
Imaging from visibilities: Maximum Entropy Method (MEM)
CountCount visibilityvisibility spectraspectra
Spatial frequencies(ui
(j),vi(j))
Re(V(ui(j),vi
(j),q1)) Im(V(ui(j),vi
(j),q1))
Re(V(ui(j),vi
(j),qN)) Im(V(ui(j),vi
(j),qN))
• For each detector and each (u,v) point:
� construct the count visibility spectrum (count visibility vs count energy)
Electron Electron visibilityvisibility spectraspectra
Spatial frequencies(ui
(j),vi(j))
Re(W(ui(j),vi
(j),E1)) Im(W(ui(j),vi
(j),E1))
Re(W(ui(j),vi
(j),EM)) Im(W(ui(j),vi
(j),EM))
∫+∞
=q
ji
ji
ji
ji dEEqK EvuWqvuV ),();,();,( )()()()(
Electron energy range 10 – 90 keV
� apply regularized inversion to obtain an electron visibility spectrum(electron visibility vs count energy)
Electron visibilities reach energies higher than photon visibilities(thanks to bremsstrahlung)
Electron Electron visibilitiesvisibilities
Detector 1 (u1
(1),v1(1)) Re(W(u1
(1),v1(1);E)) Im(W(u1
(1),v1(1);E))
(u32(1),v32
(1)) Re(W(u32(1),v32
(1);E)) Im(W(u32(1),v32
(1);E))
Detector 9 (u1
(9),v1(9)) Re(W(u1
(9),v1(9);E)) Im(W(u1
(9),v1(9);E))
(u32(9),v32
(9)) Re(W(u32(9),v32
(9);E)) Im(W(u32(9),v32
(9);E))
Fixed energy channel [E,E+∆E]
VisibilityVisibility--basedbased electron electron mapsmaps
10-14 keV 14-18 keV 18-22 keV 22-26 keV 26-30 keV
30-34 keV 34-38 keV 38-42 keV 42-46 keV 46-50 keV
Imaging from visibilities: Maximum Entropy Method (MEM)
VisibilityVisibility--basedbased electron electron mapsmaps
50-54 keV 54-58 keV 58-62 keV 62-66 keV 66-70 keV
70-74 keV 74-78 keV 78-82 keV 82-86 keV 86-90 keV
Imaging from visibilities: Maximum Entropy Method (MEM)
10-14 keV 14-18 keV 18-22 keV 22-26 keV 26-30 keV
PhotonPhoton mapsmaps vsvs electron electron mapsmaps
30-34 keV 34-38 keV 38-42 keV 42-46 keV 46-50 keV
PhotonPhoton mapsmaps vsvs electron electron mapsmaps
50-54 keV 54-58 keV 58-62 keV 62-66 keV 66-70 keV
PhotonPhoton mapsmaps vsvs electron electron mapsmaps
70-74 keV 74-78 keV 78-82 keV 82-86 keV
PhotonPhoton mapsmaps vsvs electron electron mapsmaps