The relative timing of RHESSI and radio Phoenix2 fine structures

1
The relative timing of RHESSI and radio
(Phoenix-2) fine structures

• Kaspar Arzner and Arnold Benz,
• PSI / ETHZ

• RHESSI Demodulation Method
• Benchmarking
• HXR-Radio comparison procedure
• Results

4th RHESSI Workshop, Meudon, July 26-28 2004
2
Motivation

• Physical emission mechanisms supposedly known
  HXR by Bremsstrahlung type III from electron
  beams -gt Langmuir waves -gt electromagnetic waves
  etc.
• But same electrons?
• But geometry? Radio propagation delay or
  electron time-of-flight delay? Center-to-limb
  variation?
• Earlier observations have reported HXR
  finestructures down to some 50ms, and different
  HXR/radio timings.
• Is there a canonical flare at all?
• What can RHESSI contribute? Good time resolution,
  but on time scales lt 2s, RHESSI must be
  demodulated before comparison with spatially
  non-resolved radio data from the Phoenix-2
  instrument.

3
______________Visibility-based RHESSI
DemodulatPrinciple (1)
4
__________Visibility-based RHESSI Demodulation,
Principle (2)
5
__________Visibility-based RHESSI Demodulation,
Principle (3)
Poisson likelihood
A priori probability Pa
6
Technicalities
__________________Visibility-based RHESSI
Demodulation (4)

• Time bins Dt are integer fractions of the spin
  period
• Explicitly, log L S-litcit(1lnlit/cit)
• log Ptot log L - log Pa is iteratively
  maximized (Newton-Marquardt)
• Typically 104 fit parameters
• Connection between regularisation parameters ak
  and smoothness of solution rk,t
• tk max(Dt, (ak)1/2 ltcgt/lta0Lgt )

Counts livetime (0ltLlt1)
7
Benchmarking (1)
_____________________________________
Benchmarking (1)
8
Details
_____________________________________
Benchmarking (2)

• bp hsi_bproj(), set time_range, image_dim,
  xyoffset,pixel_size, cbe_powers_of_two0,
  use_auto_time_bin0, time_bin_min, time_bin_def
• vr 0,0,....0,1,0,...,0
• HSI_MODUL_PATTERN hsi_annsec2xy(bp -gt
  getdata(vratevr, this_det5))
• a0 gridtranrel_det_eff
• a1 gridtranrel_det_effmodampl
• MANUAL MODUL PATTERN (x,y) a0
  a1cos(kx(x-Px)ky(y-Py)phi)
• im hsi_image(), set as above
• map fltarr(image_dim)
• mapi,j 1
• HSI_MODUL_PROFILE im_ob -gt getdata(classhsi_mo
  dul_profile, vimagehsi_xy2annsec(map,im),
  this_det_index5)
• MANUAL MODUL PROFILE (t) a0(t)
  a1(t)cos(kx(t)(xs-Px(t))ky(t)(ys-Py(t))phi(t)
  )

9
Running the Code ...
10
Convergence
_____________________________________
Benchmarking (3)
total prob.
Agreement with observation (likelihood)
A priori prob. (smoothness)
Data from 15-Apr-02 085110 085150
11
Simulations
_____________________________________
Benchmarking (4)
true
Demod 1-9 Avg 1-3 and t 0.1s
Similar to c) but includes data gaps
Demod 7-9 Avg 7-9 and t 0.25s
12
_____________________________ Demodulation
Examples (1)
1.2 ct / Dt background found
Dt 0.013 s
13
_____________________________ Demodulation
Examples (2)
Predicted vs observed counts
Visibilities
14
____________________________ Demodulation
Examples (3)
Counts should reflect a0
Check DF ltlt 2p
15
_________________________ Demodulation
Benchmarking (5)
Visibility-based demodulation using different
subsets of subcollimators
Gordon Hurfords demodulator_test
Time bin RHESSI spin period
16
_________________________ Demodulation
Benchmarking (6)
17
_________________________ Demodulation
Benchmarking (7)
18
_________________________ Demodulation
Benchmarking (8)
19
Phoenix-2 Spectrometer

• f 112 ... 3980 MHz
• Df 1,3,10 MHz
• t 440 ms (1 freq channel)
• 100 ms (1 freq sweep)
• Tsys 1200 ... 2100 K
• HPBW full-sun
• R/L polarization only intensity used for HXR
  correlation
• Bleien, Switzerland
• (-80644 E, 4702029 N )

20
basic calibration
3. Select time -frequency box from Phoenix-2
spectrogram. Include Type III onset at high
frequencies.
Dots raw counts
1. Choose RHESSI energy band and time range
2. Generate demodulation
SC1, Dt0.2s
Here a slow and a fast component
21
Cross-Correlation
22
_________________________________________ Results
(1)
INTENSITY
Normal-drifting type III
23
_________________________________________ Results
(2)
Reversed-drifting type III
24
_________________________________________ Results
(3)
dm type III (harmonic, fundamental more delayed)
25
_________________________________________ Results
(4)
26
_________________________________________ Results
(5)
Normal-drifting type III
Synchrotron
27
_________________________________________ Results
(6)
(same event, later time)
28
_________________________________________ Results
(7)
29
_________________________________________ Results
(8)
However, ...
General agreement between HXR maximum and
synchrotron (?) patch, but there are clearly
HXR peaks without any radio association.
30
Summary
Radio is typically delayed by 0.2 0.6 seconds
(compatible with results from Markus Aschwanden
et al 1995?)
31
Discussion

• Only few (lt20) flares show correlating HXR/radio
  finestructures
• Best between type III onset and HXR
• Achievable RHESSI resolution 100 ms, depends on
  count rate.
• Uncertainties of the RHESSI demodulation are
  usually dominated by systematic (non-Poisson)
  errors (in this order) short-time source
  morphology, validity of model assumption
  B(x,t)Srk(t)Bk(x) lifetime, aspect solution
  higher harmonics GRM data ...
• Phoenix calibration by Pascal StHilaires basic
  method. Good absolute Phoenix timing (GPS).
• Light travel time RHESSI/Phoenix lt RE/c20ms
• By-eye association of individual peaks gives
  similar average delay as cross-correlation of the
  full time series.
• No indication for center-to-limb increase of
  radio delay found, but sample is small.
• If we dare a physics speculation perhaps, the
  electrons are draged out of a collisional
  population, so that HXR arises on the way to
  high energies. So HXR could mark the (low)
  acceleration site, and (normally drifting type
  III) radio traces the outward esacping beams. But
  0.4s gtgt tcoll, chrom.

32
(No Transcript)
33
_________________________________________ Results
(7b)
34
_____________________________ Demodulation
Examples (1)
Error bars perturb solution until log L drops
by unity.
35
A problem case
Attenuator in
Oscillations indicate that fit is impossible
probably I treat the attenuated grid transmission
a0 incorrectly...
36
(No Transcript)
37
(No Transcript)
38
HXR leads radio type III onset by typically 0.5s
100 MHz
PHOENIX-2 Intensity
4 GHz
RHESSI
Visibility-based demodulation