Master Thesis Quantitative Analysis of Nonthermal ......Master Thesis Quantitative Analysis of...
Transcript of Master Thesis Quantitative Analysis of Nonthermal ......Master Thesis Quantitative Analysis of...
Master Thesis
Quantitative Analysis of Nonthermal
Components in Solar Hard X-ray Flares
Observed with the Yohkoh Satellite
「ようこう」衛星で観測された硬X線
太陽フレアにおける非熱的成分の定量解析
Takashi Minoshima
簑島 敬
Department of Earth and Planetary Science
Graduate School of Science, The University of Tokyo
Superviser : Takaaki Yokoyama
January, 2005
Abstract
The observed X-ray emission within a deka-keV energy range (called “Hard X-ray
(HXR)“) associated with a solar flare is thought to be evidence that deka-keV
nonthermal electrons are efficiently produced in solar flares. However, the particle
acceleration process in solar flares is still unknown. The analysis of solar hard
X-ray flares will be useful for discussing the particle acceleration process in solar
flares. In this thesis, we discuss characteristics of nonthermal components in solar
flares by analyzing solar hard X-ray flares observed with Yohkoh quantitatively
and statistically.
To discuss characteristics of nonthermal components in solar flares quanti-
tatively, the determination of the lower energy cutoff (Ec) in the spectrum of
nonthermal electrons is very important. However, it is not easy to derive it from
the observed HXR spectrum. Consequently, we try to derive it by assuming the
energy balance between nonthermal components and thermal components in the
impulsive phase. We apply this method to seven impulsive flare events observed
with Yohkoh, and we successfully estimate physical variables in flares such as the
Ec.
The values of the derived Ec are ranging in 20 - 45 keV. The validity of these
indirectly derived Ec is roughly provided from the HXR spectral analysis. We
suggest positive correlation between the nonthermal electron rate in the impul-
sive phase and the number density of the SXR emitting flare plasma in the pre-
impulsive phase. Positive correlation between the derived Ec and the spatial scale
of the flare is also suggested. It is the first time to our knowledge that such rela-
tionships are quantitatively shown. We expect that our work will be of benefit to
the understanding of the particle acceleration process in solar flares.
i
Contents
Abstract i
1 General Introduction 1
2 Instrumentation 9
2.1 Hard X-ray Telescope (HXT) . . . . . . . . . . . . . . . . . . 9
2.1.1 Collimator (HXT-C) . . . . . . . . . . . . . . . . . . . 12
2.1.2 Detector Assembly (HXT-S) . . . . . . . . . . . . . . . 12
2.1.3 Electronics Unit (HXT-E) . . . . . . . . . . . . . . . . 12
2.2 Soft X-ray Telescope (SXT) . . . . . . . . . . . . . . . . . . . 13
2.3 Data Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3 Hard X-ray Emission Models in a Solar Flare 15
3.1 Nonthermal Bremsstrahlung . . . . . . . . . . . . . . . . . . . 16
3.1.1 Thin-Target Emission Model . . . . . . . . . . . . . . . 17
3.1.2 Thick-Target Emission Model . . . . . . . . . . . . . . 18
3.1.3 Thick-Target Emission by the Power-Law Electrons
with a Lower Energy Cutoff . . . . . . . . . . . . . . . 19
3.2 Thermal Bremsstrahlung . . . . . . . . . . . . . . . . . . . . . 22
3.3 Summary of This Chapter . . . . . . . . . . . . . . . . . . . . 24
4 Quantitative Analysis of Nonthermal Components in Solar
Hard X-ray Flares 25
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
iii
iv Contents
4.2 Analysis Method . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.2.1 Total Energy of Nonthermal Electrons . . . . . . . . . 27
4.2.2 SXR Emitting Plasma Energy . . . . . . . . . . . . . . 27
4.2.3 Energy Ratio . . . . . . . . . . . . . . . . . . . . . . . 28
4.3 Observational Data . . . . . . . . . . . . . . . . . . . . . . . . 29
4.4 Event Selection . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.5 Analysis Example (2001/04/12 X2.0 Flare) . . . . . . . . . . . 30
4.6 Results of the Statistical Analysis . . . . . . . . . . . . . . . . 35
5 Summary & Discussion 39
5.1 HXR Spectral Analysis in the Rising Phase . . . . . . . . . . . 40
5.2 Nonthermal Electron Rate . . . . . . . . . . . . . . . . . . . . 45
5.3 Lower Energy Cutoff in the Spectrum of Nonthermal Electrons 46
5.4 Event Selection: Revisited . . . . . . . . . . . . . . . . . . . . 49
5.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . 49
A Case Studies 53
Acknowledgement 67
List of Figures
1.1 Soft X-ray image of a solar flare observed on 1992 February
21 with Yohkoh/SXT. . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Soft X-ray cusp shaped structure observed with Yohkoh/SXT
(Tsuneta et al., 1992). . . . . . . . . . . . . . . . . . . . . . . 3
1.3 An example of plasmoid ejection observed with Yohkoh/SXT
(Ohyama & Shibata, 1998). . . . . . . . . . . . . . . . . . . . 3
1.4 Loop top hard X-ray source observed in the 1992 January 13
flare. Background color is soft X-ray image observed with
Yohkoh/SXT. Contour image is hard X-ray image observed
with Yohkoh/HXT. White thick line represents the solar limb
(Masuda et al., 1994). . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Cartoon of the so-called CHSKP model. Solid lines show the
magnetic field lines. Red arrows show the directions of plasma
flow. Reconnected magnetic field lines form the flaring loops
with a cusp shaped structure (red regions). Along the re-
connected lines, nonthermal particles and thermal conduction
propagate into the chromosphere, and generate “Hα flare rib-
bons” (blue regions). . . . . . . . . . . . . . . . . . . . . . . . 5
1.6 Time profile of a solar flare (Kane, 1974). . . . . . . . . . . . . 7
1.7 Energy spectra of HXR emission observed in the 1980 June 27
flare (Lin et al., 1981). . . . . . . . . . . . . . . . . . . . . . . 8
2.1 Yohkoh image. . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
v
vi List of Figures
2.2 Schematic drawing of the HXT instrument (Kosugi et al.,
1991). It consists of three major sections: the collimator
(HXT-C), the detector assembly (HXT-S), and the electron-
ics unit (HXT-E). The aspect system (HXA) is installed along
the central axis of HXT-C and HXT-S. . . . . . . . . . . . . . 11
3.1 Calculated HXR spectra emitted by the power-law electrons
with a lower energy cutoff in both “a sharp cutoff” (red aster-
isk) and “a saturation” (blue plus sign) cases. No significant
difference between them can be seen. . . . . . . . . . . . . . . 21
3.2 Calculated HXR spectrum emitted by the power-law electrons
with a sharp lower energy cutoff (asterisk) and the fitted dou-
ble power-law spectrum (red line). Fitting parameters are
εb = 28.2 keV, γ1 = 2.27, and γ2 = 3.98, respectively. . . . . . 22
4.1 HXR time profiles observed with Yohkoh/HXT in the 2001
April 12 flare. . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.2 Hard X-ray image (contour) taken with the HXT M2-band,
overlaid on a soft X-ray image in the 2001 April 12 flare.
Hard X-ray image is synthesized by the MEM. The photon
count accumulation interval for hard X-ray image is 10:16:32 -
10:17:48 UT (impulsive phase). Contour levels are 70, 35, 12.5
% of the maximum brightness. Soft X-ray image is taken with
SXT (Be119) at 10:17:11 UT. Heliographic grids are shown by
dashed lines in 2◦ increments. . . . . . . . . . . . . . . . . . . 32
4.3 Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma
(blue asterisk) in the 2001 April 12 flare. . . . . . . . . . . . . 33
4.4 Plot of |Enonth(Ec)−∆Eth|/∆Eth as a function of Ec. . . . . . 34
4.5 Plot of the derived Ec vs. M2-band peak count rate. When
the derived Ec is greater(lower) than 30 keV, called “relatively
high(low)“. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
List of Figures vii
4.6 Plot of the estimated nonthermal electron rate in the impulsive
phase vs. the volume emission measure (left panel), the num-
ber density (right panel) of the SXR emitting flare plasma in
the pre-impulsive phase. Correlation coefficients are 0.94 and
0.97, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.1 HXR time profiles observed with Yohkoh/HXT in the 2001
April 6 flare. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.2 Plot of the single power-law photon index obtained from the
M1-band and L-band count data (horizontal axis) vs. that
obtained from the M2-band and M1-band count data (vertical
axis) in the rising phase of the 2001 April 6 flare. . . . . . . . 42
5.3 HXR time profiles observed with Yohkoh/HXT in the 2000
June 2 flare. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.4 Plot of the single power-law photon index obtained from the
M1-band and L-band count data (horizontal axis) vs. that
obtained from the M2-band and M1-band count data (vertical
axis) in the rising phase of the 2000 June 2 flare. . . . . . . . . 44
5.5 Plot of the estimated nonthermal electron rate in the impulsive
phase vs. the number density of the SXR emitting flare plasma
in the pre-impulsive phase (same as the right panel of Figure
4.6). Correlation coefficient is 0.97. Solid line is the power-law
fit by the least square method. . . . . . . . . . . . . . . . . . . 46
5.6 Hard X-ray image taken with the HXT M2-band in the 2001
August 25 flare. The two HXR sources are specified by eye
(the boxed areas). . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.7 Plot of the derived Ec vs. the calculated footpoint distance.
Correlation coefficient is 0.8. Solid line is the linear fit by the
least square method. . . . . . . . . . . . . . . . . . . . . . . . 48
A.1 HXR time profiles observed with Yohkoh/HXT in the 1997
November 6 flare. . . . . . . . . . . . . . . . . . . . . . . . . . 57
viii List of Figures
A.2 Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma
(blue asterisk) in the 1997 November 6 flare. . . . . . . . . . . 57
A.3 HXR time profiles observed with Yohkoh/HXT in the 1998
August 18 flare. . . . . . . . . . . . . . . . . . . . . . . . . . . 58
A.4 Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma
(blue asterisk) in the 1998 August 18 flare. . . . . . . . . . . . 58
A.5 Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma
(blue asterisk) in the 2000 June 2 flare. . . . . . . . . . . . . . 59
A.6 HXR time profiles observed with Yohkoh/HXT in the 2000
November 24 flare. . . . . . . . . . . . . . . . . . . . . . . . . 60
A.7 Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma
(blue asterisk) in the 2000 November 24 flare. . . . . . . . . . 60
A.8 Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma
(blue asterisk) in the 2001 April 6 flare. . . . . . . . . . . . . . 61
A.9 HXR time profiles observed with Yohkoh/HXT in the 2001
August 25 flare. . . . . . . . . . . . . . . . . . . . . . . . . . . 62
A.10 Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma
(blue asterisk) in the 2001 August 25 flare. . . . . . . . . . . . 62
List of Tables
4.1 Selected seven impulsive flares observed with Yohkoh. . . . . . 30
4.2 Results of the statistical analysis. . . . . . . . . . . . . . . . . 36
5.1 Physical variables discussed in Section 5.1through Section 5.3. 49
ix
Chapter 1
General Introduction
The solar atmosphere is full of dynamic plasma phenomena. Recent obser-
vations have revealed that the solar atmosphere is much dynamic than we
thought, even if the Sun is “quiet”. Above all, a solar flare (Figure 1.1)
is the largest exploding phenomenon in our solar system. Its spatial scale
reaches 105 km, which is 10 times larger than the radius of the Earth. Its
energy reaches 1032 erg, which is 100 million times larger than the energy
released by a hydrogen bomb. Once the flare occurs, a huge amount of en-
ergy is released in a short time, and solar plasma is strongly heated and
accelerated. Electromagnetic waves over a very wide range of wavelengths
from X-rays (occasionally from γ-rays) to radio waves are emitted during the
flare. Almost all flares are accompanied by a huge amount of mass ejection,
the so-called coronal mass ejections (CMEs), into interplanetary space. It
affects the Earth’s magnetosphere, yields geomagnetic storm or aurora.
Classically, solar flares were merely recognized as sudden exploding and
brightening phenomena on the photosphere. But recent notable progress of
observational instruments such as Yohkoh (Ogawara et al., 1991), SOHO1,
and TRACE2 has made it possible to study the physics of solar flares in
detail. The magnetic reconnection model is widely accepted to explain the
1http://sohowww.nascom.nasa.gov/2http://vestige.lmsal.com/TRACE/
1
2 1. General Introduction
Figure 1.1: Soft X-ray image of a solar flare observed on 1992 February 21
with Yohkoh/SXT.
mechanism of a solar flare. The magnetic reconnection, topological change of
a configuration of antiparallel magnetic field lines in the presence of plasma, is
a fundamental physical process and is thought to be a highly efficient engine
to convert magnetic energy into thermal and kinetic energies of plasma flows
and particles. Observed phenomena associated with solar flares, such as cusp
shaped structure (Tsuneta et al., 1992, Figure 1.2), plasmoid ejections (e.g.,
Ohyama & Shibata, 1998, Figure 1.3), loop-top hard X-ray sources (Masuda
et al., 1994, Figure 1.4), reconnection inflow (Yokoyama et al., 2001), and so
on, morphologically confirm the standard magnetic reconnection model, the
so-called CSHKP model (Carmichael, 1964; Sturrock, 1966; Hirayama, 1974;
Kopp & Pneuman, 1976, Figure 1.5).
3
Figure 1.2: Soft X-ray cusp shaped structure observed with Yohkoh/SXT
(Tsuneta et al., 1992).
Figure 1.3: An example of plasmoid ejection observed with Yohkoh/SXT
(Ohyama & Shibata, 1998).
4 1. General Introduction
Figure 1.4: Loop top hard X-ray source observed in the 1992 January 13 flare.
Background color is soft X-ray image observed with Yohkoh/SXT. Contour
image is hard X-ray image observed with Yohkoh/HXT. White thick line
represents the solar limb (Masuda et al., 1994).
Solar flares are often observed in X-ray range. The interesting is that
X-ray emissions within a few times keV energy range (the so-called “Soft
X-ray (SXR)“) and within a deka-keV energy range (the so-called “Hard
X-ray (HXR)“) show quite different characteristics. Light curve of SXR
often shows “gradual” rise and decay, whereas that of HXR often shows
“impulsive” rise and decay (Figure 1.6). Energy spectra of SXR and HXR
also show different characteristic. Energy spectrum of SXR is well-fitted by
a (single temperature) Maxwellian, whereas that of HXR is well-fitted by a
(single or double) power-law function (e.g., Lin et al., 1981, Figure 1.7). The
observed power-law energy spectrum within HXR energy range indicates the
existence of deka-keV nonthermal particles in solar flares.
We can also see the spatial difference between SXR and HXR emission
sites. In SXR observation, the brilliant loop-like structure (a flare loop) is
5
Figure 1.5: Cartoon of the so-called CHSKP model. Solid lines show the
magnetic field lines. Red arrows show the directions of plasma flow. Re-
connected magnetic field lines form the flaring loops with a cusp shaped
structure (red regions). Along the reconnected lines, nonthermal particles
and thermal conduction propagate into the chromosphere, and generate “Hα
flare ribbons” (blue regions).
often seen. In HXR observation, on the other hand, we can often see the
discrete HXR sources located at the footpoints of a flare loop (See Figure
1.4 ). Sakao (1994) concluded that the high energy nonthermal electrons
precipitation into the dense chromosphere along a flare loop produce HXR
radiation via bremsstrahlung.
As mentioned above, it is the observational fact that high energy nonther-
mal particles are efficiently produced in solar flares. However, the particle
acceleration process is still unknown. The presence of a loop-top HXR source,
which is one of the most important findings made by Yohkoh, implies that the
acceleration site is located above a flare loop(s). But further observational
results and analyses are needed to confirm this proposition. Some theoretical
6 1. General Introduction
models about the particle acceleration mechanism in solar flares have been
discussed, but none of them are conclusive. Thus the particle acceleration
process in solar flares is one of the most important scientific issues in solar
physics.
In this thesis, we discuss characteristics of nonthermal components in so-
lar flares by analyzing hard X-ray flares observed with Yohkoh quantitatively
and statistically. We successfully estimate physical variables in flares such
as the lower energy cutoff in the spectrum of nonthermal electrons. Fur-
thermore, we examine relationships of physical variables of flare nonthermal
components. We expect that this work will be of benefit to the understanding
of the particle acceleration process in solar flares.
The plan of this thesis is as follows. In Chapter 2 we briefly review
instrumentation used for our analysis. In Chapter 3 we review hard X-ray
emission models applied to the observed hard X-ray emission. In Chapter 4
we detail our analysis. In Chapter 5 we present summary and discussion.
7
Figure 1.6: Time profile of a solar flare (Kane, 1974).
8 1. General Introduction
Figure 1.7: Energy spectra of HXR emission observed in the 1980 June 27
flare (Lin et al., 1981).
Chapter 2
Instrumentation
In this Chapter, we briefly review instrumentation used for our analysis. We
carry out the analysis by using the observational data of the Yohkoh satellite
(Ogawara et al., 1991, Figure 2.1) and the GOES satellite. The Yohkoh
satellite is a project of the Institute of Space and Astronautical Sciences,
Japan (ISAS1) and an international collaboration with the US and UK. It
was launched on 1991 August 30, and operated for more than a decade until
2001 December 14. Yohkoh carried four scientific instruments: the Hard X-
ray Telescope (HXT; Kosugi et al., 1991), the Soft X-ray Telescope (SXT;
Tsuneta et al., 1991), the Wide Band Spectrometer (WBS), and the Bragg
Crystal Spectrometer (BCS). The HXT and SXT were so designed as to
obtain simultaneous hard and soft X-ray flare images with high spatial and
temporal resolution. These instruments are powerful imagers to study the
high energy physics of solar flares.
2.1 Hard X-ray Telescope (HXT)
The hard X-ray telescope (HXT) is a Fourier-synthesis type imager with
64 bi-grid modulation subcollimators (SC’s). Each SC measures a spatially
modulated photon count. A set of photon count data from the 64 SC’s
1Now, Japan Aerospace Exploration Agency (JAXA)
9
10 2. Instrumentation
Figure 2.1: Yohkoh image.
can be converted into an image by using image-synthesis procedures such as
Maximum Entropy Method (MEM).
The HXT records a set of 64 photon counts when the flare mode is trig-
gered. The main capabilities of HXT are as follows:
i) Simultaneous imaging in four energy bands, namely, the L-band (13.9 -
22.7 keV), M1-band (22.7 - 32.7 keV), M2-band (32.7 - 52.7 keV), and H-
band (52.7 - 92.8 keV);
ii) Angular resolution of ∼ 5′′ with a wide field of view covering the whole
Sun (∼ 35′ by 35′);
iii) Basic temporal resolution of 0.5 s; and
iv) High sensitivity with an effective area of ∼ 60 cm2.
Instrumentally HXT consists of three major sections, i.e., the collimator
(HXT-C), the detector assembly (HXT-S), and the electronics unit (HXT-
E) (Figure 2.2). Outlines of HXT-C, HXT-S, and HXT-E are described in
Section 2.1.1 through 2.1.3, respectively.
2.1. Hard X-ray Telescope (HXT) 11
Figure 2.2: Schematic drawing of the HXT instrument (Kosugi et al., 1991).
It consists of three major sections: the collimator (HXT-C), the detector
assembly (HXT-S), and the electronics unit (HXT-E). The aspect system
(HXA) is installed along the central axis of HXT-C and HXT-S.
12 2. Instrumentation
2.1.1 Collimator (HXT-C)
The collimator (HXT-C) is the X-ray optics part of the instrument. It is
a metering tube (417 mm × 376 mm × 1400 mm) with X-ray grid plates
at both ends. Each grid plate is an assembly of 64 subcollimator grids,
which provides the 64 modulation patterns necessary for image-synthesis.
The aspect system (HXA) optics, which include lenses with filters on the
front grid plate and fiducial marks on the rear plate, is equipped at the
center of the X-ray optics. It provides the optical axis direction information
of the X-ray optics with respect to the solar disk.
2.1.2 Detector Assembly (HXT-S)
The detector assembly is composed of 64 detector modules, eight high-voltage
power supply units, and two one-dimensional CCD arrays for aspect optics.
A detector module mainly consists of a NaI(T1) scintillation crystal and a
photomultiplier tube with a high-voltage bleeder string and a pre-amplifier
installed. Eight detector modules are packed together to form a detector unit
to which one DC-DC converter supplies high voltage. Eight detector units
are tied together to form the detector assembly.
2.1.3 Electronics Unit (HXT-E)
The electronics unit (HXT-E) processes the signals from HXT-S. It converts
hard X-ray pulse-height analogue signals from the individual subcollimators
into digital signals and counts the incident photon number after discriminat-
ing the photon energy into four energy bands. All digitalized photon count
data are sent to an onboard data processor (DP; Ogawara et al., 1991) every
0.5 s. HXT-E also processes the HXA signals. In addition, HXT-E controls
power/mode of the whole HXT instrument.
For more details, see Kosugi et al. (1991). We use the HXT data to
estimate physical variables of nonthermal components in solar flares (Chapter
2.2. Soft X-ray Telescope (SXT) 13
4).
2.2 Soft X-ray Telescope (SXT)
The soft X-ray telescope (SXT) is a grazing-incidence reflection telescope
which forms X-ray images in the 0.25 - 4.0 keV range on a CCD detector
(1024 × 1024 pixels). A filter wheel assembly located in front of the CCD
detector provides the capability of energy discrimination for plasma temper-
ature and emission measure diagnostics. A 119 µm beryllium filter (Be119)
and a 11.6 µm Al filter (thick Al) are sensitive to the highest energies. A
rotating shutter, also located in front of the CCD detector, provides the ca-
pability to optimize exposure time in response to the solar activity. Both are
automatically controlled by an onboard data processor. The highest spatial
resolution is ∼ 2.5′′/pixel in full-resolution images in the partial frame image
mode (64 × 64 pixels). Temporal resolution is usually up to 2 s.
We use the SXT data to estimate the volume of flare region (Chapter 4).
2.3 Data Set
In this thesis, we use the observational data of Yohkoh/HXT, Yohkoh/SXT,
and GOES. The HXT and SXT data are obtained from PLAIN Center
DARTS2. Flare events suitable for our analysis are selected by surveying
the whole flares detected with HXT, which are listed in “The YOHKOH
HXT/SXT Flare Catalogue” (Sato et al., 2003). We select seven impulsive
flare events (see Section 4.4 for more details).
The HXT data is a set of photon count data from the 64 SC’s every
0.5 s and in four energy bands, as mentioned in Section 2.1. We apply
background subtraction and, if necessary, photon counts correction of the
HXT data “saturated” due to scalar overflow and/or detector deadtime to
the data used for our analysis.
2http://www.darts.isas.ac.jp/index.html
14 2. Instrumentation
We use the SXT data in full-resolution images in the partial frame image
mode (flare mode). Filter Be119 is selected because this is suitable to an-
alyze the high-temperature phenomena such as solar flares. The SXT data
used for our analysis are correctly calibrated by using standard calibrating
procedure. When the SXT images contain a saturated pixel(s), these images
are eliminated from our analysis.
GOES is the broadband soft X-ray telescope operated by NOAA. It ob-
serves whole Sun X-ray fluxes in the 0.5 - 4.0 and 1.0 - 8.0 A wavelength
bands (no spatial resolution except for GOES-12 SXI). We use the GOES
data for plasma temperature and emission measure diagnostics by taking the
ratio of emission observed with these two bands. Note that all amount of X-
ray fluxes detected with GOES is assumed to be emitted by the flare thermal
plasma which SXT (Be119) observes simultaneously.
Chapter 3
Hard X-ray Emission Models in
a Solar Flare
In this Chapter, we review hard X-ray (HXR) emission models applied to the
observed HXR emission in a solar flare. A significant amount of HXRs, not
only soft X-rays (SXRs), is emitted during a flare. Main characteristics of the
observed HXRs, which are quite different from those of the observed SXRs,
are (1) light curve which shows impulsive fluctuation in the early phase of
a flare (the so-called “impulsive phase”); and (2) energy spectrum which is
well-fitted by a power-law function. These characteristics of the observed
HXRs indicate that a large amount of nonthermal particles is impulsively
produced in the impulsive phase. Moreover, it is generally accepted that a
significant amount of flare energies is released during the impulsive phase.
This means that nonthermal particles play an important role in flare energet-
ics. Therefore, the study of flare nonthermal particles is important to discuss
not only the particle acceleration process in solar flares but also the nature
of flares themselves. To do this, HXRs will be a powerful mean.
HXR emission in a solar flare is a continuum emission. It is widely be-
lieved that interaction of beams of nonthermal electrons with ambient so-
lar plasma results in HXR emission via bremsstrahlung, i.e., nonthermal
bremsstrahlung. We first review nonthermal bremsstrahlung in Section 3.1.
15
16 3. Hard X-ray Emission Models in a Solar Flare
3.1 Nonthermal Bremsstrahlung
Two major models of nonthermal bremsstrahlung are the thin-target and
thick-target emission models proposed by Brown (1971). In the thin-target
case it is assumed that nonthermal electrons continue to emit HXRs without
significant modification to their distribution function. In the thick-target
case, on the other hand, it is assumed that nonthermal electrons are im-
mediately stopped (thermalized) by Coulomb collisions with dense ambient
plasma. A thin-target scenario would be applicable to nonthermal electrons
injected outward through the corona, or trapped into low-density corona. A
thick-target scenario is applicable to nonthermal electrons precipitating into
the dense chromosphere.
HXR emission in a solar flare is in the deka-keV energy range. In such a
energy range, cross section for the bremsstrahlung is expressed as a conve-
nient formula. It is direction-integrated, nonrelativistic, Bethe-Heitler cross
section (Jackson, 1962):
σB(ε, E) =8α
3r20
mec2
εEln
1 + (1− ε/E)1/2
1− (1− ε/E)1/2
=κBH
εEln
1 + (1− ε/E)1/2
1− (1− ε/E)1/2(3.1)
=7.9× 10−25
εEln
1 + (1− ε/E)1/2
1− (1− ε/E)1/2[cm2keV−1].
Here α = e2/(hc) is the fine structure constant, r0 = e2/(mec2) is the classical
electron radius, κBH = (8α/3)r20mec
2 is the constant in the Bethe-Heitler
cross section, ε is photon energy in units of keV, E is electron energy in units
of keV, respectively. The number of photons emitted by bremsstrahlung per
unit time per unit energy per unit volume is expressed as follows:
1
ε
dW
dεdV dt= npneve(E)σB(ε, E). (3.2)
Here np is the target density, ne is the source electron density, and ve(E) is
the velocity of electrons, respectively. Equation (3.2) can be rewritten as
1
ε
dW
dεdt=
∫
V
dV · (npneveσB). (3.3)
3.1. Nonthermal Bremsstrahlung 17
This equation expresses the number of HXR photons emitted from the whole
bremsstrahlung sites per unit time per unit energy. The HXR spectrum
I(ε) (photons/cm2/sec/keV) observed at the orbit of the Earth is simply
written as
I(ε) =1
4πR2
(1
ε
dW
dεdt
)
=1
4πR2
∫
V
dV · (npneveσB) [photons/cm2/sec/keV], (3.4)
where R = 1AU.
3.1.1 Thin-Target Emission Model
Let us consider the thin-target case. In this case, it is assumed that the
distribution of source electrons is not modified throughout HXR emission.
Thus, for a power-law spectrum of source electrons:
F (E) = AE−δ [electrons/sec/keV], (3.5)
right hand side of equation (3.3) becomes∫
V
dV · (npneveσB) = A∆N
∫ ∞
ε
E−δσB(ε, E)dE, (3.6)
where ∆N =∫source
np(s)ds is the column density of the source observed.
Thus, the HXR spectrum (eq. (3.4)) in the thin-target case is written as
Ithin(ε) =∆NAκBH
4πR2ε−1
∫ ∞
ε
E−(δ+1) ln1 + (1− ε/E)1/2
1− (1− ε/E)1/2dE, (3.7)
by using equation (3.1). This equation can be simplified by evaluating the
integral by parts (see, e.g., Tandberg-Hanssen & Emslie, 1988):
Ithin(ε) =∆NAκBH
4πR2
B(δ, 1/2)
δε−(δ+1) (3.8)
∝ ε−(δ+1) [photons/cm2/sec/keV].
Here B(a, b) is the standard beta function. Under the thin-target assumption,
we find that if the source electron spectrum is a power-law, then the resultant
18 3. Hard X-ray Emission Models in a Solar Flare
HXR spectrum is also a power-law Ithin(ε) = aε−γ, and the spectral index
of the HXR spectrum is larger than that of the source electron spectrum
(γ = δ + 1).
3.1.2 Thick-Target Emission Model
Let us consider the thick-target case. In this case, it is assumed that non-
thermal electrons immediately lose their energy via Coulomb collisions with
dense ambient plasma. Thus, we first summarize expressions for Coulomb
collisions. The energy loss rate by Coulomb collisions is expresses as follows:
dE
dt= −σE(E)npve(E)E. (3.9)
Here σE(E) is the cross section for the Coulomb energy loss and has the
following form:
σE(E) =2πe4 ln Λ
E2=
K
E2, (3.10)
where ln Λ is the Coulomb logarithm which is treated as almost constant.
Suppose that an electron with initial energy E0 collides with dense target
plasma and emits photons by bremsstrahlung until it is thermalized. The
number of photons per unit energy ν(ε, E0), at an energy of ε, emitted in
above situation is expressed as
ν(ε, E0) =
∫ end(E=ε)
start(E=E0)
npσB(ε, E)dl
=
∫ t1(E=ε)
t0(E=E0)
npσB(ε, E(t))ve(t)dt
=
∫ E0
ε
σB(ε, E)
EσE(E)dE, (3.11)
where we use equation (3.9). By using this equation, right hand side of
equation (3.3) becomes∫
V
dV · (npneveσB) =
∫
S
dS
∫
L
dl · (npσBneve)
=
∫
S
dS
∫ E0
ε
dEσB(ε, E)
EσE(E)neve. (3.12)
3.1. Nonthermal Bremsstrahlung 19
Thus, for a power-law spectrum of injected electrons:
F (E0) = AE−δ0 [electrons/sec/keV], (3.13)
the HXR spectrum (eq. (3.4)) in the thick-target case is written as
Ithick(ε) =1
4πR2
∫ E0=∞
E0=ε
F (E0)
∫ E=E0
E=ε
σB(ε, E)
EσE(E)dEdE0
=AκBH
4πR2Kε−1
∫ E0=∞
E0=ε
E−δ0
∫ E=E0
E=ε
ln1 + (1− ε/E)1/2
1− (1− ε/E)1/2dEdE0,
(3.14)
by using equations (3.1) and (3.10). This equation can be simplified by re-
versing the order of integration considering the area of integration in (E, E0)
space (see also, e.g., Tandberg-Hanssen & Emslie, 1988):
Ithick(ε) =AκBH
4πR2K
B(δ − 2, 1/2)
(δ − 1)(δ − 2)ε−(δ−1) (3.15)
∝ ε−(δ−1) [photons/cm2/sec/keV].
Also under the thick-target assumption, we find that if the electron injection
spectrum is a power-law, then the resultant HXR spectrum is also a power-
law Ithick(ε) = aε−γ. The spectral index of the HXR spectrum is smaller than
that of the electron injection spectrum (γ = δ − 1).
To summarize, we can numerically derive the electron injection spectrum
F (E0) = AE−δ0 from the observed power-law HXR spectrum Ithick(ε) = aε−γ
under the thick-target assumption as follows (Hudson et al., 1978):
A = 3.28× 1033γ2(γ − 1)2B
(γ − 1
2,3
2
)· a,
δ = γ + 1.
(3.16)
3.1.3 Thick-Target Emission by the Power-Law Elec-
trons with a Lower Energy Cutoff
In both the thin-target and thick-target emission models reviewed in Sections
3.1.1 and 3.1.2 respectively, a power-law spectrum of HXR emitting electrons
20 3. Hard X-ray Emission Models in a Solar Flare
extends to lower energies indefinitely (cf. eq. (3.5) and (3.13)), that is, the
number of nonthermal electrons becomes infinity. Therefore, the spectrum
of nonthermal electrons must have a lower energy cutoff (Ec) to keep the
number of nonthermal electrons within a reasonable range. It seems to be
known that when the spectrum of nonthermal electrons has the lower energy
cutoff, the resultant HXR spectrum becomes flat toward the lower energy
and the broken energy of HXR spectrum (εb) is smaller than Ec (e.g., Nitta
et al., 1990). Gan et al. (2001b) studied this issue quantitatively.
In this Section, we present the “modified” thick-target emission model
based on Gan et al. (2001b). The HXR spectrum in the thick-target case is
written as (cf. eq. (3.14))
Ithick(ε) = I0ε−1
∫ E0=∞
E0=ε
F (E0)
∫ E=E0
E=ε
ln1 + (1− ε/E)1/2
1− (1− ε/E)1/2dEdE0, (3.17)
where I0 = κBH/(4πR2K). Here F (E0) will be given as a power-law with a
lower energy cutoff of the form:
F (E0) =
{AE−δ
0 , if E0 ≥ Ec,
C, if E0 < Ec,(3.18)
instead of equation (3.13). Here we consider C = 0 (“a sharp cutoff” case)
or C = AE−δc (“a saturation” case). Putting equation (3.18) into equation
(3.17), we get the modified thick-target emission form:
if ε ≥ Ec,
Ithick(ε) = AI0ε−1
∫ E0=∞
E0=ε
E−δ0
∫ E=E0
E=ε
ln1 + (1− ε/E)1/2
1− (1− ε/E)1/2dEdE0,
(3.19)
otherwise,
Ithick(ε) = CI0ε−1
∫ E0=Ec
E0=ε
∫ E=E0
E=ε
ln1 + (1− ε/E)1/2
1− (1− ε/E)1/2dEdE0
+ AI0ε−1
∫ E0=∞
E0=Ec
E−δ0
∫ E=E0
E=ε
ln1 + (1− ε/E)1/2
1− (1− ε/E)1/2dEdE0.
(3.20)
3.1. Nonthermal Bremsstrahlung 21
Figure 3.1: Calculated HXR spectra emitted by the power-law electrons
with a lower energy cutoff in both “a sharp cutoff” (red asterisk) and “a
saturation” (blue plus sign) cases. No significant difference between them
can be seen.
Obviously, equation (3.19) is same as equation (3.14). This means that
equation (3.16) is applicable to the HXR spectrum with energy above Ec. We
numerically calculate equations (3.19) and (3.20) in both “a sharp cutoff”
and “a saturation” cases. The resultant HXR spectra in case of Ec = 40 keV
and δ = 5 are shown in Figure 3.1.
These calculated HXR spectra can be well-fitted by a double power-law
of the form:
I(ε) =
{a1ε
−γ1 , if ε < εb,
a2ε−γ2 , if ε ≥ εb,
(3.21)
where a2 = a1εγ2−γ1
b . The condition γ1 < γ2 is said to “break down” (Dulk
et al., 1992). Figure 3.2 shows the calculated HXR spectrum in case of
Ec = 40 keV, δ = 5, and “a sharp cutoff”, and the fitted double power-law
22 3. Hard X-ray Emission Models in a Solar Flare
Figure 3.2: Calculated HXR spectrum emitted by the power-law electrons
with a sharp lower energy cutoff (asterisk) and the fitted double power-law
spectrum (red line). Fitting parameters are εb = 28.2 keV, γ1 = 2.27, and
γ2 = 3.98, respectively.
spectrum by using equation (3.21). Gan et al. (2001b) studied a theoretical
relationship between γ1 and γ2, as well as the relationship between εb/Ec and
γ2, to establish how to determine the lower energy cutoff from the observed
HXR spectrum. We do not show details of the theoretical results of Gan
et al. (2001b), but note the rough relationship of the following form:
εb = (0.6− 0.8)Ec. (3.22)
3.2 Thermal Bremsstrahlung
It is well-known that the observed HXR spectrum is often characterized by
a combination of exponential-like and power-law rather than a single power-
law (e.g., Lin et al., 1981). It is also well-known that the observed SXR
3.2. Thermal Bremsstrahlung 23
spectrum is well-fitted by a single temperature Maxwellian (although several
line emissions contaminate the SXR continuum spectrum). Such exponential-
like spectrum is interpreted as thermal bremsstrahlung from the (super) hot
plasma. In this Section, we briefly summarize thermal bremsstrahlung.
Let us consider a volume V , containing a uniform density ne of hot elec-
trons with an isothermal Maxwellian energy distribution, viz.
f(E) = 2πne
(1
πkT
)3/2
E1/2e−E/kT [electrons/cm3/erg], (3.23)
where k is the Boltzmann’s constant. The bremsstrahlung produced by in-
teraction of these electrons with ambient plasma is written as
Ithermal(ε) =npV
4πR2
∫ ∞
ε
f(E)ve(E)σB(ε, E)dE
=1
4πR2
(8
πme
)1/2npneV
(kT )3/2
κBH
ε
×∫ ∞
ε
exp(−E/kT ) ln1 + (1− ε/E)1/2
1− (1− ε/E)1/2dE (3.24)
by using equation (3.4). Integrating by parts, we find
Ithermal(ε) =1
4πR2
(8κ2
BH
πmek
)1/2npneV
εT 1/2exp(−ε/kT )g(ε/kT ), (3.25)
where
g(a) =
∫ ∞
0
e−ax
√x(1 + x)
dx (3.26)
is a slowly varying function of order unity. We may neglect this factor. Thus,
equation (3.25) is written as
Ithermal(ε) =D
4πR2
EM
εT 1/2exp(−ε/kT ) [photons/cm2/sec/keV], (3.27)
where D =√
(8k2BH/πmek) = 5.7 × 10−12 cm3s−1K1/2, and EM = npneV '
n2eV is the volume emission measure of the source.
To summarize, we can deduce the volume emission measure and the tem-
perature of the thermal bremsstrahlung source from the observed exponential-
like X-ray spectrum.
24 3. Hard X-ray Emission Models in a Solar Flare
3.3 Summary of This Chapter
In this Chapter, we reviewed several HXR emission models applied to the
observed HXR emission in a solar flare. In Section 3.1, we reviewed nonther-
mal bremsstrahlung. Two major models of nonthermal bremsstrahlung are
the thin-target and thick-target emission models.
In the thin-target emission model reviewed in Section 3.1.1, we find that if
the source electron spectrum is a power-law F (E) = AE−δ, then the resultant
HXR spectrum is also a power-law Ithin(ε) = aε−γ, and the spectral index
of the HXR spectrum is larger than that of the source electron spectrum
(γ = δ + 1).
In the thick-target emission model reviewed in Section 3.1.2, we find that
if the electron injection spectrum is a power-law F (E0) = AE−δ0 , then the
resultant HXR spectrum is also a power-law Ithick(ε) = aε−γ, and the spectral
index of the HXR spectrum is smaller than that of the electron injection
spectrum (γ = δ − 1). Using equation (3.16), we can numerically derive the
electron injection spectrum from the observed power-law HXR spectrum.
In Section 3.1.3, we presented the modified thick-target emission model by
considering the lower energy cutoff in the spectrum of nonthermal electrons.
When the spectrum of nonthermal electrons has the lower energy cutoff,
the resultant HXR spectrum becomes flat toward the lower energy and the
broken energy of HXR spectrum is smaller than the lower energy cutoff in
the spectrum of nonthermal electrons. We show that, in the modified thick-
target emission model, the numerical relationship between (A, δ) and (a, γ)
(eq. (3.16)) is applicable to the HXR spectrum with energy above Ec. Rough
relationship between εb and Ec is given by equation (3.22).
In Section 3.2, we reviewed thermal bremsstrahlung. We find that the
volume emission measure and the temperature of the thermal bremsstrahlung
source can be deduced from the observed exponential-like X-ray spectrum,
such as the SXR spectrum.
Chapter 4
Quantitative Analysis of
Nonthermal Components in
Solar Hard X-ray Flares
4.1 Introduction
Many observations have revealed that a significant amount of nonthermal
particles is produced in a solar flare and nonthermal particles play an im-
portant role in flare energetics. However, the particle acceleration process is
still unknown. Therefore, the study of flare nonthermal particles is very im-
portant. Because it is widely believed that the flare hard X-rays are emitted
by nonthermal electrons, the analysis of solar hard X-ray flares can provide
information about nonthermal electrons in solar flares, and it may provide
an explanation about the acceleration process in solar flares. Aim of this
thesis is to discuss characteristics of nonthermal components in solar flares
by analyzing hard X-ray flares quantitatively and statistically.
To discuss characteristics of nonthermal components in solar flares quan-
titatively, the determination of the lower energy cutoff (Ec) in the spectrum
of nonthermal electrons is very important. It determines the total number
and energy of nonthermal electrons, and it may be related to the acceleration
25
26 4. Quantitative Analysis of Nonthermal Components in Solar Hard X-ray Flares
mechanism in solar flares. Because HXR spectrum becomes flat toward the
lower energy and the broken energy of HXR spectrum (εb) is smaller than
Ec when the spectrum of nonthermal electrons has the lower energy cutoff,
to estimate the broken energy of the observed HXR spectrum must be a key
clue to deduce the lower energy cutoff in the spectrum of nonthermal elec-
trons. However, it is not easy to obtain it from the observed HXR spectrum
because of contamination due to thermal emission at the lower energy. For
a long time, one assumes the Ec to be 20 keV or 30 keV without any justifi-
cation. Quantitative determination of the lower energy cutoff is necessary to
discuss characteristics of nonthermal components. Consequently, we try to
derive the lower energy cutoff in the spectrum of nonthermal electrons with
some assumptions.
4.2 Analysis Method
It is well-known that the variation of the time integral of the microwave
intensity Imicro often closely matched the variation of the SXR intensity (ISXR)
in solar flares, the so-called “Neupert effect” (Neupert, 1968):
Imicro ∝ dISXR
dt. (4.1)
Because the variation of the HXR intensity (IHXR) is similar to that of the
microwave intensity, the concept of the Neupert effect can be extended to
include the relationship between the SXR and the HXR intensity, viz.
IHXR ∝ dISXR
dt. (4.2)
Since the SXR intensity is thought to represent the total energy of the SXR
emitting flare plasma and the HXR intensity would be thought to represent
the energy release rate of nonthermal electrons, the concept of the Neupert
effect will be additionally extended as follows:∫
dEnonth
dtdt = α∆ESXR. (4.3)
4.2. Analysis Method 27
Here dEnonth/dt is the energy rate of nonthermal electrons, ∆ESXR is the
energy build-up of the SXR emitting flare plasma during the impulsive phase,
and α represents the ratio between the total energy of nonthermal electrons
and the energy build-up of the SXR emitting flare plasma, respectively. In
the Section 4.2.1 through 4.2.3, we will show how to derive the lower energy
cutoff in the spectrum of nonthermal electrons from equation (4.3).
4.2.1 Total Energy of Nonthermal Electrons
Adopting the thick-target emission model, we can numerically derive the
spectrum of nonthermal electrons F (E) = AE−δ from the observed HXR
spectrum and equation (3.16). A and δ are obtained as a function of time,
namely, A(t) and δ(t). The energy rate of nonthermal electrons is given as
follows:
dEnonth
dt=
∫ ∞
Ec
EA(t)E−δ(t)dE
=A(t)
δ(t)− 2E−δ(t)+2
c . (4.4)
In general, Ec must be a function of time. But we assume that it is time-
independent. The total energy of nonthermal electrons released during the
impulsive phase (left hand side of eq. (4.3)) is given as follows:∫
dEnonth
dtdt =
∫A(t)
δ(t)− 2E−δ(t)+2
c dt ≡ Enonth(Ec). (4.5)
Because both A(t) and δ(t) are given, equation (4.5) is a function of Ec.
4.2.2 SXR Emitting Plasma Energy
Based on Wu et al. (1986), we evaluate the energy of the SXR emitting flare
plasma. ESXR is expressed as follows:
ESXR = Eth + Econvect + Etr + Erad + Econduct, (4.6)
where Eth is the thermal energy, Econvect is the convection energy, Etr is the
turbulent energy, Erad is the time integral of the radiation loss rate, and
28 4. Quantitative Analysis of Nonthermal Components in Solar Hard X-ray Flares
Econduct is the time integral of the thermal conduction loss rate of the SXR
emitting flare plasma, respectively. In general, Econvect and Etr are 2-3 orders
of magnitude lower than Eth. Typical time scale of the radiation loss rate
of the SXR emitting flare plasma (T ∼ 107 K) is an order of magnitude
longer than the impulsive duration (∼ 100 sec). Therefore, Econvect, Etr,
and Erad are generally negligible compared with Eth during the impulsive
phase. Evaluation of Econduct, on the other hand, is difficult because it is
very sensitive to temperature. Consequently, we will neglect this term in
this thesis. We simply evaluate ESXR as follows:
ESXR ' Eth = (3nekT )V = 4.14× 10−16EM1/2V 1/2T, (4.7)
where EM is the volume emission measure, V is the volume, and T is the
temperature of the SXR emitting flare plasma, respectively. Since all of these
variables are estimated from the SXR observational data, ∆ESXR ' ∆Eth is
calculable.
4.2.3 Energy Ratio
The ratio between the total energy of nonthermal electrons and the energy
build-up of the SXR emitting flare plasma is important quantity in flare en-
ergetics. In our analysis, however, α is assumed to be unity. The assumption
of α = 1 means that the energy build-up of the SXR emitting flare plasma
completely results from the thermalization of nonthermal electrons.
Hence, we can numerically derive the lower energy cutoff in the spectrum
of nonthermal electrons by solving the following equation:
Enonth(Ec)−∆Eth = 0. (4.8)
Using the derived Ec, we can estimate several physical variables of nonther-
mal electrons.
4.3. Observational Data 29
4.3 Observational Data
In this thesis, we use the observational data of Yohkoh/HXT, Yohkoh/SXT,
and GOES (see Section 2.3).
The Yohkoh/HXT observational data is applied to obtain the incident
HXR spectrum. Taking the ratio of the measured photon counts in two
adjacent energy bands, we can obtain the incident HXR spectral parameters,
a(t) and γ(t), with a single power-law assumption. The M2-band (32.7 - 52.7
keV) count data and the H-band (52.7 - 92.8 keV) count data are mainly
applied to obtain them in this thesis.
The Yohkoh/SXT (Be119) observational data is applied to estimate the
volume of the SXR emitting flare plasma. Because SXT is the 2-dimensional
soft X-ray imager, we estimate the volume of the SXR emitting flare plasma
from the SXT (Be119) observational data with simple geometric uncertainty
as follows:
V = (0.5− 1.0)× S3/2, (4.9)
where S is the area observed with SXT (Be119). The area is calculated from
the SXT pixels at which the data number exceeds 3 percent of the maximum
one in each image.
The GOES observational data is applied to obtain the volume emission
measure and the temperature of the SXR emitting flare plasma. We assume
that all amount of X-ray fluxes detected with GOES is emitted by the flare
plasma which SXT (Be119) observes simultaneously, although GOES detects
whole Sun X-ray fluxes. With this simplification, the volume emission mea-
sure and the temperature obtained from the GOES observational data are
treated as those of the SXR emitting flare plasma.
4.4 Event Selection
In this thesis, we analyze seven impulsive flares observed with Yohkoh which
are listed in Table 4.1. These flares are selected according to the following
30 4. Quantitative Analysis of Nonthermal Components in Solar Hard X-ray Flares
Table 4.1: Selected seven impulsive flares observed with Yohkoh.
M2-band peak count rate
date GOES class (cts/sec/SC)
1997/11/06 X9.4 5901
1998/08/18 X2.8 115
2000/06/02 M7.6 78
2000/11/24 X2.3 1479
2001/04/06 X5.6 992
2001/04/12 X2.0 152
2001/08/25 X5.3 3653
criteria:
(a) M2-band peak count rate exceeds 30 cts/sec/SC;
(b) Double-source structure is seen in M2-band;
(c) SXT (Be119) images without saturated pixels at the end of the HXR
burst are available; and
(d) There are both the HXT and SXT observational data in the “pre-impulsive
phase”, in which no significant enhancement of X-ray emission still occurs.
Discrimination of double-source structure is made by eye with “The YOHKOH
HXT/SXT Flare Catalogue” (Sato et al., 2003).
4.5 Analysis Example (2001/04/12 X2.0 Flare)
Let us present an analysis example. The flare commenced on 12 April,
2001 at ∼ 10:11 UT. Figure 4.1 shows the HXR time profiles observed with
Yohkoh/HXT. The GOES soft X-ray class was X2.0. Figure 4.2 shows the
X-ray images taken with HXT and SXT.
4.5. Analysis Example (2001/04/12 X2.0 Flare) 31
Figure 4.1: HXR time profiles observed with Yohkoh/HXT in the 2001 April
12 flare.
32 4. Quantitative Analysis of Nonthermal Components in Solar Hard X-ray Flares
Figure 4.2: Hard X-ray image (contour) taken with the HXT M2-band, over-
laid on a soft X-ray image in the 2001 April 12 flare. Hard X-ray image is
synthesized by the MEM. The photon count accumulation interval for hard
X-ray image is 10:16:32 - 10:17:48 UT (impulsive phase). Contour levels are
70, 35, 12.5 % of the maximum brightness. Soft X-ray image is taken with
SXT (Be119) at 10:17:11 UT. Heliographic grids are shown by dashed lines
in 2◦ increments.
4.5. Analysis Example (2001/04/12 X2.0 Flare) 33
∆Eth
∆tT05 T90
Figure 4.3: Time profiles of the M2-band count rate (red line) and the es-
timated thermal energy of the SXR emitting flare plasma (blue asterisk) in
the 2001 April 12 flare.
Figure 4.3 shows the time profiles of the M2-band count rate (red line)
and the estimated thermal energy of the SXR emitting flare plasma (blue
asterisk). Black solid lines at ∼ 10:16 UT and at ∼ 10:23 UT denote “T05”
and “T90”, respectively. Here T05(T90) is the time at which the time integral
of the M2-band count rate reaches 5(90) percent of the total amount of the
measured photon counts in M2-band and is defined as the onset(end) of the
impulsive phase. The region surrounded by green lines is defined as the
pre-impulsive phase (tpre = T0 ∼ T0.05). The SXT images taken between ∼10:19 and ∼ 10:23 UT are eliminated from this analysis because they contain
a saturated pixel(s). Note that the volume of the SXR emitting flare plasma
is assumed to be V = S3/2 (see eq. (4.9)) in this example.
We first estimate the thermal energy build-up of the SXR emitting flare
34 4. Quantitative Analysis of Nonthermal Components in Solar Hard X-ray Flares
Figure 4.4: Plot of |Enonth(Ec)−∆Eth|/∆Eth as a function of Ec.
plasma ∆Eth = Eth(T90)−Eth(tpre) and the impulsive duration ∆t = T90−T05. In this example, ∆Eth of 1.3 × 1031 (erg) and ∆t of 410 (sec) are
obtained. We also estimate several physical variables in the pre-impulsive
phase such as the volume emission measure, the temperature, and the number
density n = (EM/V )1/2. The total energy of nonthermal electrons released
during the impulsive phase, Enonth(Ec), is obtained by the method mentioned
in Section 4.2.1.
Next, we derive the optimal Ec by comparing Enonth(Ec) with ∆Eth (see
Figure 4.4). In this example, the optimal Ec is derived to be 25 keV.
Finally, we estimate several physical variables of nonthermal components.
The total number of nonthermal electrons in the impulsive phase is given as
N =
∫∫ ∞
Ec
A(t)E−δ(t)dEdt, (4.10)
and the nonthermal electron rate in the impulsive phase is estimated as
4.6. Results of the Statistical Analysis 35
follows:
dN
dt∼ N
∆t. (4.11)
In this example, N and dN/dt are estimated to be 2.3× 1038 (electrons) and
5.6 × 1035 (electrons/sec), respectively. The analysis results in other flares
are shown in Appendix A.
4.6 Results of the Statistical Analysis
The results of the statistical analysis are summarized in Table 4.2 and Figure
4.5.
36 4. Quantitative Analysis of Nonthermal Components in Solar Hard X-ray Flares
Tab
le4.
2:R
esult
sof
the
stat
isti
calan
alysi
s.
EM
an
aT
a∆
Eth
∆t
Ec
NdN
/dt
M2-
band
peak
coun
tra
te
date
(104
8cm
−3)
(109
cm−
3)
(107
K)
(103
1er
g)(s
ec)
(keV
)(1
038)
(103
5se
c−1)
(cts
/sec
/SC
)
1997
/11/
060.
52.
2−
4.7
1.4
1.4−
2.0
158
>50
b···
···
5901
1998
/08/
181.
63.
0−
4.3
1.2
0.7−
1.0
406
31−
34c
1.0−
1.5
2.4−
3.7
115
2000
/06/
0220
18−
261.
30.
2−
0.3
5020−
22d
0.5−
0.7
9.0−
1478
2000
/11/
2432
28−
391.
21.
0−
1.3
146
23−
25d
1.7−
2.5
12−
1714
79
2001
/04/
062.
04.
3−
6.0
1.3
1.6−
2.3
374
43−
48c
1.6−
2.7
4.2−
7.3
992
2001
/04/
121.
63.
5−
5.0
1.3
0.9−
1.3
410
25−
28d
1.4−
2.3
3.5−
5.6
152
2001
/08/
255.
05.
5−
7.8
1.1
2.0−
2.8
464
38−
44c
2.0−
3.2
4.1−
6.9
3653
aV
aria
bles
inth
epr
e-im
puls
ive
phas
e.bT
hede
rive
dE
cha
san
unce
rtai
nty.
cT
hede
rive
dE
cis
“rel
ativ
ely
high
”.dT
hede
rive
dE
cis
“rel
ativ
ely
low
”.
4.6. Results of the Statistical Analysis 37
Figure 4.5: Plot of the derived Ec vs. M2-band peak count rate. When the
derived Ec is greater(lower) than 30 keV, called “relatively high(low)“.
We classify the analyzed events based on the derived Ec into “relatively
high Ec” ones (Ec > 30 keV) and “relatively low Ec” ones (Ec < 30 keV).
Since the derived Ec in the 1997 November 6 flare is high compared with the
energy range of M2-band (32.7 - 52.7 keV), it will have an uncertainty (see
Section 3.1.3). We don’t use this result for the following discussions in this
thesis.
From the results of the statistical analysis, we find a relationship between
the nonthermal electron rate in the impulsive phase and the volume emission
measure/the number density of the SXR emitting flare plasma in the pre-
impulsive phase. Figure 4.6 is the plot of the estimated nonthermal electron
rate in the impulsive phase vs. the volume emission measure (left panel),
the number density (right panel) of the SXR emitting flare plasma in the
pre-impulsive phase. Correlation coefficients are 0.94 and 0.97, respectively.
This result, indicating the positive correlation between the production rate
38 4. Quantitative Analysis of Nonthermal Components in Solar Hard X-ray Flares
Figure 4.6: Plot of the estimated nonthermal electron rate in the impulsive
phase vs. the volume emission measure (left panel), the number density
(right panel) of the SXR emitting flare plasma in the pre-impulsive phase.
Correlation coefficients are 0.94 and 0.97, respectively.
of nonthermal electrons and the number density of the ambient plasma in
the pre-flare stage, is thought to be plausible. It is the first time to our
knowledge that the clear relationship between nonthermal component and
thermal component is quantitatively shown.
Further discussions will be presented in Chapter 5.
Chapter 5
Summary & Discussion
In the previous Chapter, we analyzed solar hard X-ray flares observed with
Yohkoh quantitatively and statistically. Main results are as follows:
i) The values of the derived lower energy cutoff (Ec) in the spectrum of
nonthermal electrons are ranging in 20 - 45 keV; and
ii) Positive correlation between the nonthermal electron rate in the im-
pulsive phase and the volume emission measure/the number density of
the SXR emitting flare plasma in the pre-impulsive phase is shown.
The values of the derived Ec, which are slightly higher than the usually ac-
cepted 20 keV, seem to be in reasonable range, although several authors ar-
gued that the lower energy cutoff is much higher than 20 keV (e.g., Gan et al.,
2001a). In our analysis, however, the lower energy cutoff in the spectrum of
nonthermal electrons is indirectly derived by assuming the energy balance
between nonthermal components and thermal components (eq. (4.3)). The
verification of the validity of the derived Ec is necessary to confirm our anal-
ysis results. In the next Section, we will check the validity of the derived
Ec.
39
40 5. Summary & Discussion
5.1 HXR Spectral Analysis in the Rising Phase
The observed HXR spectrum is often contaminated by thermal emission at
the lower energy, that is, the nonthermal characteristic is contaminated by
the thermal characteristic. However, the HXR spectrum in the rising phase
is probably less contaminated by thermal emission because thermal emission
in the rising phase is not so intense. Therefore, we expect that the HXR
spectrum in the rising phase can directly provide information about nonther-
mal components, that is, the broken energy of HXR spectrum (εb; see Section
3.1.3) can be seen in the rising phase. Because the broken energy of HXR
spectrum is a key clue to deduce the lower energy cutoff in the spectrum of
nonthermal electrons, the analysis of the HXR spectrum in the rising phase
will be useful for discussing the validity of the derived Ec. Low spectral
resolution of Yohkoh/HXT with only four energy bands makes it difficult
to analyze HXR spectrum in detail, but rough discussion about the broken
energy of HXR spectrum is possible. In this Section, we discuss the validity
of the derived Ec by analyzing the HXR spectrum in the rising phase.
5.1. HXR Spectral Analysis in the Rising Phase 41
Figure 5.1: HXR time profiles observed with Yohkoh/HXT in the 2001 April
6 flare.
Let us first present the analysis of the HXR spectrum in the rising phase of
the 2001 April 6 flare. This flare commenced on 6 April 2001 at ∼ 19:12 UT.
Figure 5.1 shows the HXR time profiles observed with Yohkoh/HXT. The
GOES soft X-ray class was X5.6. The derived Ec is 43 - 48 keV (“relatively
high”; see Table 4.2). Figure 5.2 shows the relationship between the single
power-law index obtained from the M1-band and L-band count data, and that
obtained from the M2-band and M1-band count data in the rising phase of
this flare. This result means that the observed HXR spectrum in the rising
phase of this flare shows the broken-down spectral form, like Figure 3.1 or
3.2. Using the measured photon counts in HXT four energy bands, we can
obtain the HXR spectral parameters with a double power-law (eq. (3.21))
assumption (double power-law fitting). From the double power-law fitting,
the broken energy of the HXR spectrum in the rising phase of this flare
42 5. Summary & Discussion
Figure 5.2: Plot of the single power-law photon index obtained from the M1-
band and L-band count data (horizontal axis) vs. that obtained from the
M2-band and M1-band count data (vertical axis) in the rising phase of the
2001 April 6 flare.
is derived to be ∼ 28 keV. The observed HXR spectra in the rising phase
of other “relatively high Ec” flares show the tendency similar to this. The
values of the broken energy of the HXR spectrum in the rising phase of these
flares are derived to be ∼ 25 keV. Since the relationship between εb and Ec is
expressed by equation (3.22), the tendency that the derived Ec is “relatively
high” (Ec > 30 keV) is not inconsistent with the derived εb of 25 - 28 keV.
5.1. HXR Spectral Analysis in the Rising Phase 43
Figure 5.3: HXR time profiles observed with Yohkoh/HXT in the 2000 June
2 flare.
Next, we present the analysis of the HXR spectrum in the rising phase of
the 2000 June 2 flare. This flare commenced on 2 June 2000 at ∼ 18:52 UT.
Figure 5.3 shows the HXR time profiles observed with Yohkoh/HXT. The
GOES soft X-ray class was M7.6. The derived Ec is 20 - 22 keV (“relatively
low”; see Table 4.2). Figure 5.4 shows the relationship between the single
power-law index obtained from the M1-band and L-band count data, and
that obtained from the M2-band and M1-band count data in the rising phase
of this flare. Figure 5.4 is quite different from Figure 5.2. We can not see
the broken-down HXR spectral form in the rising phase of this flare. The
observed HXR spectra in the rising phase of other “relatively low Ec” flares
show the tendency similar to this. This result will be interpreted as follows: If
the lower energy cutoff in the spectrum of nonthermal electrons is “relatively
low” (Ec < 30 keV), then the broken energy of the resultant HXR spectrum
44 5. Summary & Discussion
Figure 5.4: Plot of the single power-law photon index obtained from the M1-
band and L-band count data (horizontal axis) vs. that obtained from the
M2-band and M1-band count data (vertical axis) in the rising phase of the
2000 June 2 flare.
is estimated to be εb <∼ 20 keV from equation (3.22). Such εb is in the L-band
energy range (13.9 - 22.7 keV). Therefore, it will be hard to identify this from
the HXR spectral analysis with HXT.
The discussion in this Section is summarized as follows:
(a) The observed HXR spectra in the rising phase of the “relatively high
Ec” flares show the broken-down spectral form and the values of the broken
energy of the HXR spectrum in the rising phase of these flares are 25 - 28
keV;
5.2. Nonthermal Electron Rate 45
(b) The observed HXR spectra in the rising phase of the “relatively low Ec”
flares, on the other hand, don’t show the broken-down spectral form.
These tendencies are not inconsistent with a theoretical relationship by
Gan et al. (2001b). Thus, we conclude that the validity of the lower en-
ergy cutoff in the spectrum of nonthermal electrons derived in the previous
Chapter is roughly provided.
5.2 Nonthermal Electron Rate
In this Section, dependence of the nonthermal electron rate on the number
density of the SXR emitting flare plasma in the pre-impulsive phase is dis-
cussed. Figure 5.5 shows the relationship between the estimated nonthermal
electron rate in the impulsive phase (dN/dt) and the number density of the
SXR emitting flare plasma in the pre-impulsive phase (n). Correlation coeffi-
cient is 0.97. The power-law fit by the least square method is dN/dt ∝ n0.61.
In the 2-dimensional magnetic reconnection model, the electron injection rate
into the reconnection region is roughly evaluated as the product of the mass
flux into the reconnection region 2nvin and the area of the reconnection re-
gion Srec. By assuming the nonthermal electron rate to be proportional to
the electron injection rate into the reconnection region, it is simply evaluated
as follows:
dN
dt∝ nvinSrec
≈ nMAvASrec ∝ MABcSrecn1/2, (5.1)
where vin is the inflow velocity into the reconnection region, vA is the Alfven
velocity, MA = vin/vA is the reconnection rate, and Bc is the magnetic field
strength in the corona, respectively. Our result of the power-law fit dN/dt ∝n0.61 to the observation is similar to equation (5.1) in terms of n. Therefore,
the correlation in Figure 5.5 might indicate the magnetic reconnection process
itself. However, it is very rough discussion. Above all, the values of the
number density of the SXR emitting plasma in the pre-impulsive phase have
46 5. Summary & Discussion
Figure 5.5: Plot of the estimated nonthermal electron rate in the impul-
sive phase vs. the number density of the SXR emitting flare plasma in the
pre-impulsive phase (same as the right panel of Figure 4.6). Correlation
coefficient is 0.97. Solid line is the power-law fit by the least square method.
a strong uncertainty due to the estimation of the volume of the SXR emitting
plasma in the pre-impulsive phase (n = (EM/V )1/2). It is, however, probable
that the nonthermal electron rate in the impulsive phase and the number
density of the SXR emitting plasma in the pre-impulsive phase have the
positive correlation.
5.3 Lower Energy Cutoff in the Spectrum of
Nonthermal Electrons
In this Section, dependence of the lower energy cutoff in the spectrum of
nonthermal electrons on the spatial scale of the flare is discussed. To reveal
5.3. Lower Energy Cutoff in the Spectrum of Nonthermal Electrons 47
Figure 5.6: Hard X-ray image taken with the HXT M2-band in the 2001
August 25 flare. The two HXR sources are specified by eye (the boxed
areas).
this dependence will be useful for understanding the acceleration mechanism
in solar flares. Because all the analyzed flare events show typical double-
source structure in M2-band (Section 4.4), we evaluate the spatial scale of
these flares by the HXR sources separation distance (the footpoint distance)
in HXT M2-band image at the HXR peak time.
To measure the footpoint distance, we first specify the two HXR sources
by eye (Figure 5.6) and determine the centroids of the HXR sources (xi, yi)
in image coordinate. Next, we convert the image coordinates (xi, yi) into
the heliographic coordinates (li, bi) (see Aschwanden et al., 1999b, Appendix
A, for the heliographic coordinate transformations). Finally, we calculate
48 5. Summary & Discussion
Figure 5.7: Plot of the derived Ec vs. the calculated footpoint distance.
Correlation coefficient is 0.8. Solid line is the linear fit by the least square
method.
the footpoint distance d on the solar surface as follows (Aschwanden et al.,
1999a):
d = 2πR¯
√[(l1 − l2) cos b1]2 + (b1 − b2)2
360◦, (5.2)
where R¯ = 7× 105 km ≈ 1000′′ is the solar radius. Calculation results are
shown in Table 5.1.
Figure 5.7 shows the relationship between the derived Ec and the cal-
culated footpoint distance. Correlation coefficient is 0.8. The linear fit by
the least square method is Ec = 0.98d + 2.7. This result is suggestive. The
straightforward interpretation about this result is that the DC electric field
acceleration of electrons out of a thermal plasma (e.g., Holman, 1985) effi-
ciently works in the impulsive phase of a solar flare, if the footpoint distance
corresponds to the size of the acceleration region. Further analyses are needed
5.4. Event Selection: Revisited 49
to discuss this issue, on which we will continue to study.
Table 5.1: Physical variables discussed in Section 5.1 through Section 5.3.
Ec εb d dN/dt
date (keV) (keV) (arcsec) (1035 sec−1)
1998/08/18 31− 34 ∼ 25 31.3 2.4− 3.7
2000/06/02 20− 22 · · · 18.8 9.0− 14
2000/11/24 23− 25 · · · 26.2 12− 17
2001/04/06 43− 48 ∼ 28 31.7 4.2− 7.3
2001/04/12 25− 28 · · · 26.4 3.5− 5.6
2001/08/25 38− 44 ∼ 25 42.6 4.1− 6.9
5.4 Event Selection: Revisited
In our analysis, flare events are selected according to the several criteria (see
Section 4.4). Some of them give a strict limitation against event selection.
Very impulsive flares are not selected for our analysis because of the criterion
’(c) SXT (Be119) images without saturated pixels at the end of the HXR
burst are available’. The criterion ’(d) There are both the HXT and SXT
observational data in the “pre-impulsive phase”’ also gives a strict limitation
against event selection. Because, in most flares observed with Yohkoh, the
X-ray intensity rapidly increases after the flare mode onset, this criterion
limits the number of available events. Therefore, the selected flares for our
analysis are “particular” ones. In fact, the six of the selected flares are GOES
X class flares. Our results can not be necessarily applied to other flares, such
as very impulsive flares, weak flares, and long-duration flares.
5.5 Concluding Remarks
In this thesis, we studied characteristics of nonthermal components in solar
flares by analyzing solar hard X-ray flares observed with Yohkoh quantita-
50 5. Summary & Discussion
tively and statistically. We successfully estimated physical variables in solar
flares such as the lower energy cutoff in the spectrum of nonthermal electrons
with some assumptions (Chapter 4). Relationships of physical variables of
flare nonthermal components are suggested (Chapter 5). It is the first time
to our knowledge that such relationships are quantitatively shown.
Many problems of flare physics still remain. Here we mention the two
of still remaining problems. First problem is the “total number problem”
stated, e.g., by Miller et al. (1997). This problem is that the total number
of nonthermal electrons derived from the observed HXR spectrum is larger
than the total number of electrons contained in the whole flare region. In this
thesis, this problem is remained unsolved. Because the total number of non-
thermal electrons closely depend on the lower energy cutoff in the spectrum
of nonthermal electrons (eq. (4.10)), accurate determination of the lower
energy cutoff is necessary to settle this problem. The Reuven Ramaty High
Energy Solar Spectroscopic Imager1 (RHESSI), which is the newest solar hard
X-ray/gamma-ray imager, may give us an answer to this problem because
it has an ability to observe solar hard X-rays with high spectral and spatial
resolution. Second problem is about particle acceleration process. There
remain many questions about particle acceleration process, such as accelera-
tion mechanism, acceleration timing and duration, acceleration site, and so
on. We presented a rough discussion in Section 5.3. Highly spatially-resolved
spectral analysis (called “imaging spectroscopy”) is necessary to discuss the
particle acceleration process in solar flares more in detail. In addition, the-
oretical study of particle acceleration process is also necessary. None of the
proposed acceleration mechanism in solar flares, such as the DC electric field
acceleration, the stochastic acceleration, and the shock acceleration, are con-
clusive at the present time. Both theoretical and observational study must
be done to discuss not only the particle acceleration process in solar flares
but also the nature of flares themselves.
We expect that our work will be of benefit to the understanding of the
1http://hesperia.gsfc.nasa.gov/hessi/
5.5. Concluding Remarks 51
particle acceleration process in solar flares. And we believe that, in the
future, the mysteries of this attractive phenomenon are completely revealed.
Appendix A
Case Studies
Here we briefly present the results of the quantitative analysis in the selected
flare events except that presented in Section 4.5. The analysis method is
detailed in Chapter 4.
1997/11/06 X9.4 Flare
This flare commenced on 6 November, 1997 at ∼ 11:51 UT. Figure A.1
shows the HXR time profiles observed with Yohkoh/HXT. The GOES soft
X-ray class was X9.4.
Figure A.2 shows the time profiles of the M2-band count rate (red line)
and the estimated thermal energy of the SXR emitting flare plasma (blue
asterisk). Black solid lines at ∼ 11:53 UT and at ∼ 11:55:30 UT denote
“T05” and “T90”, respectively. The region surrounded by green lines is
defined as the pre-impulsive phase (tpre = T0 ∼ T0.05). Note that the
volume of the SXR emitting flare plasma is assumed to be V = S3/2 (see eq.
(4.9)).
The estimated physical variables are as follows: ∆Eth = Eth(T90) −Eth(tpre) = 2.0×1031 (erg), ∆t = T90−T05 = 158 (sec), and Ec > 50 (keV).
Since the derived Ec is high compared with the energy range of M2-band
(32.7 - 52.7 keV), it will have an uncertainty (see Section 3.1.3). Therefore,
we don’t estimate the total number of nonthermal electrons (N) and the
53
54 A. Case Studies
nonthermal electron rate (dN/dt) in the impulsive phase.
1998/08/18 X2.8 Flare
This flare commenced on 18 August, 1998 at ∼ 8:17 UT. Figure A.3 shows
the HXR time profiles observed with Yohkoh/HXT. The GOES soft X-ray
class was X2.8.
Figure A.4 shows the time profiles of the M2-band count rate (red line)
and the estimated thermal energy of the SXR emitting flare plasma (blue
asterisk). Black solid lines at ∼ 8:18:40 UT and at ∼ 8:25:20 UT denote
“T05” and “T90”, respectively. The region surrounded by green lines is
defined as the pre-impulsive phase (tpre = T0 ∼ T0.5). Note that the volume
of the SXR emitting flare plasma is assumed to be V = S3/2.
The estimated physical variables are as follows: ∆Eth = Eth(T90) −Eth(tpre) = 1.0 × 1031 (erg), ∆t = T90 − T05 = 406 (sec), Ec = 31 (keV),
N = 1.5× 1038 (electrons), and dN/dt = 3.7× 1035 (electrons/sec).
2000/06/02 M7.6 Flare
This flare commenced on 2 June, 2000 at ∼ 18:52 UT. Figure 5.3 shows
the HXR time profiles observed with Yohkoh/HXT. The GOES soft X-ray
class was M7.6.
Figure A.5 shows the time profiles of the M2-band count rate (red line)
and the estimated thermal energy of the SXR emitting flare plasma (blue
asterisk). Black solid lines at ∼ 19:22 UT and at ∼ 19:22:50 UT denote
“T05” and “T90”, respectively. The region surrounded by green lines is
defined as the pre-impulsive phase (tpre = T0 ∼ T0.5). Note that the volume
of the SXR emitting flare plasma is assumed to be V = S3/2.
The estimated physical variables are as follows: ∆Eth = Eth(T90) −Eth(tpre) = 3.0 × 1030 (erg), ∆t = T90 − T05 = 50 (sec), Ec = 20 (keV),
N = 7.0× 1037 (electrons), and dN/dt = 1.4× 1036 (electrons/sec).
2000/11/24 X2.3 Flare
This flare commenced on 24 November, 2000 at ∼ 14:54 UT. Figure A.6
55
shows the HXR time profiles observed with Yohkoh/HXT. The GOES soft
X-ray class was X2.3.
Figure A.7 shows the time profiles of the M2-band count rate (red line)
and the estimated thermal energy of the SXR emitting flare plasma (blue
asterisk). Black solid lines at ∼ 15:08 UT and at ∼ 15:10 UT denote “T05”
and “T90”, respectively. The region surrounded by green lines is defined as
the pre-impulsive phase (tpre = T0 ∼ T0.05). Note that the volume of the
SXR emitting flare plasma is assumed to be V = S3/2.
The estimated physical variables are as follows: ∆Eth = Eth(T90) −Eth(tpre) = 1.3 × 1031 (erg), ∆t = T90 − T05 = 146 (sec), Ec = 23 (keV),
N = 2.5× 1038 (electrons), and dN/dt = 1.7× 1036 (electrons/sec).
2001/04/06 X5.6 Flare
This flare commenced on 6 April, 2001 at ∼ 19:12 UT. Figure 5.1 shows
the HXR time profiles observed with Yohkoh/HXT. The GOES soft X-ray
class was X5.6.
Figure A.8 shows the time profiles of the M2-band count rate (red line)
and the estimated thermal energy of the SXR emitting flare plasma (blue
asterisk). Black solid lines at ∼ 19:14:30 UT and at ∼ 19:20 UT denote
“T05” and “T90”, respectively. The region surrounded by green lines is
defined as the pre-impulsive phase (tpre = T0 ∼ T0.05). Note that the
volume of the SXR emitting flare plasma is assumed to be V = S3/2.
The estimated physical variables are as follows: ∆Eth = Eth(T90) −Eth(tpre) = 2.3 × 1031 (erg), ∆t = T90 − T05 = 374 (sec), Ec = 43 (keV),
N = 2.7× 1038 (electrons), and dN/dt = 7.3× 1035 (electrons/sec).
2001/08/25 X5.3 Flare
This flare commenced on 25 August, 2001 at ∼ 16:25 UT. Figure A.9 shows
the HXR time profiles observed with Yohkoh/HXT. The GOES soft X-ray
class was X5.3.
Figure A.10 shows the time profiles of the M2-band count rate (red line)
and the estimated thermal energy of the SXR emitting flare plasma (blue
56 A. Case Studies
asterisk). Black solid lines at ∼ 16:30 UT and at ∼ 16:38 UT denote “T05”
and “T90”, respectively. The region surrounded by green lines is defined as
the pre-impulsive phase (tpre = T0 ∼ T0.05). Note that the volume of the
SXR emitting flare plasma is assumed to be V = S3/2.
The estimated physical variables are as follows: ∆Eth = Eth(T90) −Eth(tpre) = 2.8 × 1031 (erg), ∆t = T90 − T05 = 464 (sec), Ec = 38 (keV),
N = 3.2× 1038 (electrons), and dN/dt = 6.9× 1035 (electrons/sec).
57
Figure A.1: HXR time profiles observed with Yohkoh/HXT in the 1997
November 6 flare.
T05 T90
∆Eth
∆t
Figure A.2: Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma (blue asterisk)
in the 1997 November 6 flare.
58 A. Case Studies
Figure A.3: HXR time profiles observed with Yohkoh/HXT in the 1998 Au-
gust 18 flare.
T90T05
∆Eth
∆t
Figure A.4: Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma (blue asterisk)
in the 1998 August 18 flare.
59
T05 T90
∆Eth
∆t
Figure A.5: Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma (blue asterisk)
in the 2000 June 2 flare.
60 A. Case Studies
Figure A.6: HXR time profiles observed with Yohkoh/HXT in the 2000
November 24 flare.
T05 T90
∆Eth
∆t
Figure A.7: Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma (blue asterisk)
in the 2000 November 24 flare.
61
T90T05
∆Eth
∆t
Figure A.8: Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma (blue asterisk)
in the 2001 April 6 flare.
62 A. Case Studies
Figure A.9: HXR time profiles observed with Yohkoh/HXT in the 2001 Au-
gust 25 flare.
T90T05 ∆Eth
∆t
Figure A.10: Time profiles of the M2-band count rate (red line) and the
estimated thermal energy of the SXR emitting flare plasma (blue asterisk)
in the 2001 August 25 flare.
Bibliography
Aschwanden, M. J., Fletcher, L., Sakao, T., Kosugi, T., & Hudson, H. 1999a,
ApJ, 517, 977
Aschwanden, M. J., Newmark, J. S., Delaboudiniere, J., Neupert, W. M.,
Klimchuk, J. A., Gary, G. A., Portier-Fozzani, F., & Zucker, A. 1999b,
ApJ, 515, 842
Brown, J. C. 1971, Sol. Phys., 18, 489
Carmichael, H. 1964, in The Physics of Solar Flares, 451
Dulk, G. A., Kiplinger, A. L., & Winglee, R. M. 1992, ApJ, 389, 756
Gan, W., Li, Y., & Chang, J. 2001a, Chinese Journal of Astronony and
Astrophysics, 1, 453
Gan, W. Q., Li, Y. P., & Chang, J. 2001b, ApJ, 552, 858
Hirayama, T. 1974, Sol. Phys., 34, 323
Holman, G. D. 1985, ApJ, 293, 584
Hudson, H. S., Canfield, R. C., & Kane, S. R. 1978, Sol. Phys., 60, 137
Jackson, J. D. 1962, Classical Electrodynamics (Classical Electrodynamics,
New York: Wiley, 1962)
Kane, S. R. 1974, in IAU Symp. 57: Coronal Disturbances, 105–141
Kopp, R. A., & Pneuman, G. W. 1976, Sol. Phys., 50, 85
63
64 Bibliography
Kosugi, T., Masuda, S., Makishima, K., Inda, M., Murakami, T., Dotani, T.,
Ogawara, Y., Sakao, T., Kai, K., & Nakajima, H. 1991, Sol. Phys., 136, 17
Lin, R. P., Schwartz, R. A., Pelling, R. M., & Hurley, K. C. 1981, ApJ, 251,
L109
Masuda, S., Kosugi, T., Hara, H., Tsuneta, S., & Ogawara, Y. 1994, Nature,
371, 495
Miller, J. A., Cargill, P. J., Emslie, A. G., Holman, G. D., Dennis, B. R.,
LaRosa, T. N., Winglee, R. M., Benka, S. G., & Tsuneta, S. 1997, J. Geo-
phys. Res., 102, 14631
Neupert, W. M. 1968, ApJ, 153, L59
Nitta, N., Dennis, B. R., & Kiplinger, A. L. 1990, ApJ, 353, 313
Ogawara, Y., Takano, T., Kato, T., Kosugi, T., Tsuneta, S., Watanabe, T.,
Kondo, I., & Uchida, Y. 1991, Sol. Phys., 136, 1
Ohyama, M., & Shibata, K. 1998, ApJ, 499, 934
Sakao, T. 1994, Ph.D. Thesis
Sato, J., Sawa, M., Yoshimura, K., Masuda, S., & Kosugi, T. 2003, The
YOHKOH HXT/SXT Flare Catalogue
Sturrock, P. A. 1966, Nature, 211, 695
Tandberg-Hanssen, E., & Emslie, A. G. 1988, The physics of solar flares
(Cambridge and New York, Cambridge University Press, 1988, 286 p.)
Tsuneta, S., Acton, L., Bruner, M., Lemen, J., Brown, W., Caravalho, R.,
Catura, R., Freeland, S., Jurcevich, B., & Owens, J. 1991, Sol. Phys., 136,
37
Tsuneta, S., Takahashi, T., Acton, L. W., Bruner, M. E., Harvey, K. L., &
Ogawara, Y. 1992, PASJ, 44, L211
Bibliography 65
Wu, S. T., de Jager, C., Dennis, B. R., Hudson, H. S., Simnett, G. M.,
Strong, K. T., Bentley, R. D., Bornmann, P. L., Bruner, M. E., Cargill,
P. J., Crannell, C. J., Doyle, J. G., Hyder, C. L., Kopp, R. A., Lemen,
J. R., Martin, S. F., Pallavicini, R., Peres, G., Serio, S., Sylwester, J., &
Veek, N. J. 1986, in Energetic Phenomena on the Sun, 5–5
Yokoyama, T., Akita, K., Morimoto, T., Inoue, K., & Newmark, J. 2001,
ApJ, 546, L69
Acknowledgement
The author would like to thank Prof. Yokoyama T., who is my supervi-
sor, for his kindhearted guidances, valuable advices, and fruitful discussions,
throughout this work. I am very pleased with that I have stimulating years
in his laboratory.
I would like to thank Prof. Terasawa T., and Hoshino M. for their precise
and fruitful comments. I am grateful to all members of the Yohkoh team.
Especially, I would like to thank Drs. Masuda S., and Sato J. for their helpful
advices and fruitful discussions. I also would like to thank Drs. Shibasaki K.,
Shimojo S., and Asai A. for their helpful and valuable advices. They gave
me stimulating experiences at Nobeyama Radio Observatory. Significant
discussions with students in the Solar-Terrestrial Physics Group are gratefully
acknowledged.
67