1. RHESSI Observations and Data Analysis Nicole Vilmer LESIA-Observatoire de Paris Tostip- October 2003

2. The (R)HESSI (Ramaty High Energy Solar Spectroscopic Imager) Experiment Scientific goals and objectives What do we know about energetic particles at the Sun from HXR/GR observations? General Description of the Instrument How to make images at X-ray/  -ray wavelengths with Rotation Modulation Collimators? How to make X-ray/  -ray spectra at high energies? How to access to data and analyse data? Time profiles Images (co-alignment) Spectra

3. RHESSI scientific goals The Sun as an efficient particle accelerator: (large fraction of the flare energy release) High Energy Solar Physics Flare energy release Particle acceleration/transport and interaction in the solar atmosphere Large Flares BUT ALSO Microflares: coronal heating?? Solar Physics special issue 210

4. Principal Investigator: Robert Lin UCB Project Manager: Peter Harvey UCB Lead System Engineer: David CurtisUCB Lead Co-Investigator: Brian Dennis GSFC Co-Investigators: Arnold Benz ETHZ Patricia Bornmann NOAA John Brown U. of Glasgow Richard Canfield Montana State U. Carol CrannellGSFC Gordon EmslieU. Alabama Huntsville Shinzo Enome Gordon HolmanGSFC, Code 682 Hugh Hudson UCB Gordon HurfordGSFC,Code 682 Takeo KosugiNAOJ Norman Madden LBNL Reuven RamatyGSFC, Code 661 Frank van BeekDelft U. Nicole VilmerParis Observatory Tycho von RosenvingeGSFC, Code 66 Alex ZehnderPSI

5. 31/01/2002 Launched on 05/02/2002

7. Solar data from 14/02/2002 Catalog of RHESSI X-ray flares from 14/02/2002 to 05/2003  8000 flares > 12 keV  Several GOES X-class flares (at least one  -ray line flare) Several X-flares May-June 2003 http://hesperia.gsfc.nasa.gov/ssw/hessi/dbase/hessi_flare_list.txt

8. Flare Accelerated Particles Particle acceleration (Where, How Fast? How Many? Which Ones?) Particle transport and interaction in the atmosphere (How do they evolve in the ambient medium?) Injection in the interplanetary medium (Where? When? Relation with flare particles?)

9. Solar X-ray/  -ray spectrum RHESSI Energy range Pion decay radiation (ions > 100 MeV/nuc) sometimes with neutrons Ultrarelativistic Electron Bremsstrahlung Thermal components Electron bremsstrahlung  -ray lines (ions > 3 MeV/nuc)

10. Direct diagnostics of energetic particles interacting in the solar atmosphere: HXR and GR continuum: ~ electrons 10 keV-~100 MeV (acceleration timescales, number and energy spectra) No imaging above 70 keV Limited spectral resolution No imaging spectroscopy  (R)HESSI Unique observation at high spectral resolution before RHESSI From Lin et al. 1981

11. Energetic Ions  -ray line spectroscopy   ions in the 1MeV/nuc -100 MeV/nuc range  narrow deexcitation  -ray line fluences  ion energy spectrum and target abundances (i.e. solar atmosphere) Broad  -ray lines  abundances of accelerated ions  2.2 MeV deuterium line: capture line after thermalization from neutrons from nuclear reactions (Share & Murphy,2000)

12.  -ray line spectroscopy before RHESSI 19 GRS/SMM /1 CGRO/OSSE flares ( Share & Murphy, 1995, 1998 )  Ion energy spectrum from Ne (1.63)/ O (6.13): power laws down to 1 MeV/nuc and ion energy content but also dependant on abundances  /p(5 flares) from Fe(0.339)/Fe(0.847). Fe (0.339) is a pure  line   /p = 0.5 3 He/ 4 He (7 flares) 0.1 to 1 ( Ramaty & al, Mandzhavidze & al ) Share & Murphy (1995) Ramaty & Mandzhavidze, 1995 O (6.13) Ne (1.63)

13. Electron/Ion Energy Contents in G GRL flares (before RHESSI) W e>20 keV and W i>1MeV/nuc 19 SMM Flares,1 OSSE, 1 GRANAT (Ramaty & Mandzhavidze, 1999) (Murphy et al, 1997, Ramaty et al, 1997) But low energy cutoffs? Better spectral resolution at X-rays electrons Low energy ions? What happens in electron-dominated events? Adapted from Ramaty & Mandzhavidze (1999) W i>1MeV/nuc for 19 SMM flares  W e>20 keV for 19 SMM flares  W i>1MeV/nuc for OSSE 4 June 1991 ‣ W i>1MeV/nuc for PHEBUS 1 June 1991

14. X/  -ray observations and acceleration processes A  dditional constraints Variability of spectra e/p in flares & from flare to flare ( electron-dominated events) Enrichment of  /p, 3 He, heavy ions (Ne,Mg, Fe) as in impulsive SEP events Variation with time of the enhancements ElectronsIons Number10 41 (>20 keV)310 35 (>30MeV) 10 36 (> 100 keV)10 32 (> 300 MeV) Acc. times~ 100 ms @100 keV < 1s @10 MeV Duration (s) 10  ~10 mn60  hour Total energy (ergs) 10 34 (> 20 keV)10 32 - 10 33 (> 1 MeV) 10 29 (> 100 keV)10 30 (> 30 MeV) Power (ergs/s) 10 32 (> 20 keV)2 10 28 (> 30 MeV) Adapted from Chupp, 1995

15. RHESSI Characteristics Imaging Angular resolution Field of View Pointing information: Solar Aspect System (SAS) Roll Angle Systems (RASs) Spectroscopy Energy range Energy resolution Fourier-transform imaging with 9 bi-grid rotating modulation collimators 2.3  to 36  depending on energy HXR 2.3  ; GRL /GR 36  Full Sun Tens of ms for basic image 2s for detailed image SAS: Sun center <1’’ RASs: roll to 1’ 3 keV to 10 MeV < 1 keV 5 keV@ 20 MeV

17. RHESSI Spectroscopy 9 bi-segmented n Germanium detectors front (1.5cm): 3 keV-250 keV rear (7.5cm): 250 keV- 17MeV 7.1 cm  8.5 cm length Cooled to < 75K  2 sets of aluminium disk attenuators (shutters) to absorb low energy photons in case of large flares (see obs summary plots) GRL spectrum simulated for HESSI for a large flare (Smith et al, 2000)

19. Instrument Data Processing Unit: Photons interacting in the GeD generate charge pulses collected and amplified by charge sensitive amplifiers This provides Counts  Front segment: 8192 energy channels from 3 keV to 2.7 MeV (0.33 keV/channel)  Rear segment: 20 keV to 17 MeV  For each photon: energy information time of arrival to 1  s with detector & live time All these information in the fits files

20. RHESSI: Spectroscopy

22. RHESSI Imaging Grids 8 pairs tungsten 1 molybdenum Pitch: 34  - 2.7mm (steps of  3 L=1.55m Fast rotation: 12-20 rpm Dynamic range:100/1 1100 uv components in 2s No modulation for>3’ but still full spectroscopic info

23. Grid 1 (2.2’’): slit and slat widths: 20 and 14  Max energy for modulation: 100 keV (1.2mm thick) One of the thickest grid (18.6mm) used to modulate photons up to 17 MeV (35’’)

24. RHESSI Imaging Angular resolution : p/2L Arrival time and energy Of each photon

25. Aspect systems: Need to know the orientation of the collimators with respect to the direction of the Sun Provided by SAS (Solar Aspect System): measurements relative to the solar limbs to ‘’ accuracy on 10ms and 2 Roll Angle Systems: a CCD RAS and a PM based version (PMTRAS Photo-Multiplier- Tube Roll Aspect System ) currently used in the software providing roll angle to ‘ accuracy several times per rotation with respect to fixed stars. It views the star field perpendicular to the Earth-Sun line and records times at which bright stars pass through the field of view.

26. 23/07/2002 X4.8 GOES: RHESSI  -ray line flare Lin et al, 2003 Images:64’’ wide At the time of flare maximum

27. Krucker et al,2003 30 –80 keV every 27s

28. X4.8 flare : 23/07/2002-X-ray spectroscopy with RHESSI Holmann et al, 2003 Piana et al, 2003 Spectre photons T=37 MK EM= 4.1 10 49 cm -3 Thin target radiation double power law Ec= 34 keV  l=1.5 Eb=129 keV  u=2.5 Inversion of the photon spectrum Electron spectrum Extrapolation above 160 keV

29. X4.8 flare : 23/07/2002 Imaging spectroscopy with RHESSI Emslie et al, 2003 N M x0.1 Sx0.01 Photon spectrum

30. Lin et al, 2003 Share et al, 2003 Spectral analysis every 20s: 6 narrow  -ray lines Electron bremsstrahlung: 2 power laws 2.77 et 2.23 > 617 keV Broad line component 511keV and 2.23 MeV lines

31. Smith et al, 2003 Redshift (0.1-0.8 %)larger than expected for a limb flare if Downward isotropic distribution if Radial B field! SMM 5 flares Same longitude Broadening 0.1-2.1% FWHM No redshift (light curve) Redshift (Heavy curve)

32. First gamma-ray images of a flare! Gamma-ray line image displaced from 20 ’’ from electron emission site!!! Interpretation? Hurford et al, 2003 + TRACE post flare loop

33. Coronal HXR sources GOES M2.5 AR 9893 AR 9893 N21 W81 large part behind the limb H  1310-1320-1332 N23 W88 SF AR 9893 < 1323- 1338 N19 W67 SF AR9901 Coronal HXR source from 13:07 UT H  8 days earlier Vilmer, Koutroumpa, Kane, Hurford, EGS

34. Comparison of RHESSI images with TRACE images at 195 Å =flare plasma at 15 MK (coalignment between EIT and Trace during the flare)  TRACE and RHESSI 12-25 keV images before 13:07 UT (no coronal HXR sources)

35. TRACE and RHESSI 12-25 keV images after 13:07 UT (coronal HXR sources with most of the time no footpoints Most energetic part of the event)

36. TRACE and RHESSI 25-50 keV images after 13:07 UT (coronal HXR sources with most of the time no footpoints Most energetic part of the event)

37. Coronal HXR sources (> 10’’ ) above the limb, displaced from the hot magnetic structures seen with TRACE? 25-50 keV predominant coronal sources above 12-25 keV sources (faint footpoints close to max) (see previous YOHKOH/HXT obs but more dynamical and more complicated fields?)

38. RHESSI & UV & Optical Observations (B. Schmieder, A. Berlicki, G. Aulanier, N. Vilmer, DPSM) 22 oct 2002 B long (NaD1, THEMIS) HXR 6-12 keV (RHESSI) Decay phase of GOES M flare

39. RHESSI & Optical Observations (B. Schmieder, A. Berlicki, G. Aulanier, N. Vilmer, DPSM) 22 oct 2002 B long (NaD1, THEMIS) I (NaD1) Éruption H  (VTT) HXR 6-12 keV (RHESSI)

40. RHESSI & SOHO JOP 136 CDS FLARE_AR 6-12keV

41. How to access and analyse data? A few addresses Data at http://hesperia.gsfc.nasa.gov/hessidata/ ftp://hercules.ethz.ch/pub/hessi/data Level 0 packets in fits files (up to 110 Mbytes) one fits file/single orbit between local midnights multiple fits files for large flares Software : sswidl (hessi)http://hesperia.gsfc.nasa http://hessi.ssl.berkeley.edu/software/ Objcct oriented software but also Graphical User Interface (GUI) A few « quicklooks »: http://sprg.ssl.berkeley.edu/~ayshih/browser/quicklook.shtml http://rhessidatacenter.ssl.berkeley.edu/ http://sprg.ssl.berkeley.edu/~ayshih/browser/grw.shtml

43. How to make light curves: the observing summary plots Look at decimation and attenuators states Need to get observing summary files hsi_obssumm_*.fits files (see hands-on

44. How to make light curves?

45. How to make images From the modulation time profiles: inverse problem of deducing the source geometry given a set of modulation profiles from different subcollimators Several image reconstruction algorithms: « back projection »: initial estimate of the image, convolution of the image with the instrumental response  sidelobes To improve the quality: CLEAN, MEM,pixon,… Not to expect the kind of images with the morphological richness of TRACE, YOHKOH/SXT, SOHO/EIT!!!

46. How to make images Back projection: equivalent to 2D inverse Fourier transform analog to radioastronomers’ dirty maps. linear process (not the case of CLEAN,MEM,…) Deduction of the source geometry given the set of observed modulation profiles from different subcollimators oriented according to the roll angles.  importance of the aspect solution!!! PMTRAS by default sometimes necessary to change to RAS (Roll Data Base still in progress) Some useful addresses: LISTING OF ROLL DATABASE GAPS > 66 SECONDS http://sprg.ssl.berkeley.edu/~ghurford/ROLL_DBASE/ROLL_DBASE_GAPS.txt Index of /hessidata/metadata/data_gap_files/daily_summary http://hesperia.gsfc.nasa.gov/hessidata/metadata/data_gap_files/daily_summary/

47. Some examples Grid 3 Grid 4 Grid 5 Grid 6 Grid 7 Grid 8 Grid 9

48. Clean: iterative algorithm developed for radio astronomy based on the assumption that the image is a superposition of point sources

49. How to compare with other images Use the synoptic archive software to get fits files from other instruments Use the mapping software of Dominic Zarro to overlay (see hands-on) !!! Some special treatments may be needed for TRACE see http://hesperia.gsgc.nasa.goc/~ptg/trace_align

50. How to make spectra? Photons interacting in the GeD generate charge pulses collected and amplified by charge sensitive amplifiers This provides Counts which are recorded in the fits files 9 bi-segmented n Germanium detectors front: 3 keV-250 keV (NOT DETECTOR 2 (7)) rear: 250 keV- 17MeV !!!Attenuators reduce the count rates in case of large flares !!!If the memory starts to fill up a decimation algorithm throws out one out of every N events in the front segment below a given energy Also indicated in the observing summary plots; Now corrected for spectroscopy.

52. How to make spectra !!! Pile up correction for large flares (also 1st order correction for spectroscopy) Pulse pile up: 2 or more « low energy photons arrive ‘simultaneously’ and cannot be distinguished from a single high energy photon. Artefact for spectroscopy but also for imaging: low energy count-rate may appear at higher energies; « ghost » low-energy sources appearing at high energies

53. How to make spectra The background issue: RHESSI is not a low background instrument. But for most events, flares are brighter than the background!! Background issues important for  -ray flares, at low energies for flares with attenuators… Sources of variation of the RHESSI background: Passes through SAA Changes of geomagnetic latitudes Electron precipitation from belts at 40-50° latitudes (appear more strongly in rear segments)

54. Spectral Data Analysis Inverse problem:how to go from a spectrum of counts per spectrometer channel (what is recorded) to spectrum of photons per energy interval incident on the spacecraft Part of the RHESSI software (SPEX spectral inversion code) Several steps: Background subtraction (count rates before and after flares, for some flares most sophisticated techniques) Generation of the Response Matrix: (hsi_srm_*.fits) How to go from photons to counts? absorption in blankets,grids,… Compton scattering in and out the detectors,… noise in the electronic,… NOT A DIAGONAL MATRIX!! Except in the HXR range below 100 keV but NOT BELOW 15 keV IF SHUTTERS!!!

55. Spectral Data Analysis Inverse problem:how to go from a spectrum of counts per spectrometer channel (what is recorded) to spectrum of photons per energy interval incident on the spacecraft Done automatically by SPEX using forward-folding once the calibartion matrix is generated Input: model photon spectrum (power-laws or Maxwellian in energy + lines) Convolution with the reponse matrix  Count spectrum Comparison with the observed one and fit by minimizing  2 !!!Input parameters. !!! The reality of the fit must be checked (are the parameters found reasonable?? Are they consistent with other observations i.e. thermal emission observed by other instruments??