Solar flare hard X-ray spikes observed by
RHESSI: a statistical study
Jianxia Cheng
Jiong Qiu, Mingde Ding, and Haimin Wang
outline




Introduction
Observations and data analysis
Properties of HXR spikes
summary
1.Introduction





Solar flare emission at sub-second was reported in hard X-ray in 70s
and 80s (van Beek et al. 1974,1976; Hoyng et al. 1976; de Jager & de
Jonger 1978; Kiplinger et al. 1983; 1984, 1989).
Kiplinger et al. (1983) found 53 out of 3000 flares produce spikes as
short as 45 ms.
These energetic flare bursts on short timescales are believed to be
nonthermal in nature, and their temporal and spectral properties place
constraints on the physical nature of the source.
Several mechanisms: dynamic magnetic reconnection (Kliem et al.,
2000); nonthermal electron injections.
Peak energy differences: time of flight; trap model (the collisional
timescale increases with the particle energy).
An example: different criteria to define a spike
It
25-100 keV
Islow
Ir
1-7 represents 7 different energy bands
Event distribution
sample
flares
Sample flares are steeper than the spike-associated flares,
suggesting that more intensive flares have in general a greater
chance to produce spikes.
HXR spikes can occur in both impulsive and gradual events.
 spkdistribution
70% of spikes
around the peak
times of the
associated flare
 spk
Time difference between spike peak time and flare start time
normalized to the flare rise time
Most of spikes
discovered in 4060 and 60-100
keV, 20% as high
as 100-300 keV.
Spikes are most probably nonthermal in nature
All these statistical results indicate that spikes are small scale
energetic events produced during the most energetic stage of
flares. In particular, they tend to occur during the rise phase of
intensive flares. On the other hand, impulsive and gradual flares
have an equal chance to produce spikes.
3.Properties of HXR spikes
Spikes duration is range from 0.2-2s with mean value of
about 1s. It is independent of energy bands. Symmetric
rise and decay phase, this is different from flares.
Nearly all spikes have harder spectral than their underlying
components. This is agree with the general scenario that flare HXR
emission exhibits a harder spectrum at emission peaks than at valley.
Time lags between
60-100 keV and 2540 kev
Low energy delay
High energy delay
The majority of events exhibit time lags shorter than
0.5s. Mean time lag is about 0.8s and -0.74s.
Low
energy
delay
High
energy
delay
About or more than 2/3 events are low energy delayed. On average,
high energy delayed spikes have a harder count spectrum than low
energy delayed events.
summary





Both impulsive and long duration flares can produce HXR spikes with nearly
equal production rates. Flares with high peak count rates are more productive in
HXR spikes.
Almost all spikes occur in the rise phase of the flares, and a large percentage,
up to 70%, of spikes are produced at or about the flare peak times.
The mean duration of spikes is about 0.9−1.0 s, independent of photon
energies. The rise and decay times of spikes are shown to be almost the same.
This differs from ordinary flares that usually have a longer decay phase
dominated by thermal emission
Most of the spikes can be detected in very high energy bands up to 100−300
keV. The HXR spectra of spikes are harder than those of the underlying slowvarying components. This fact implies the nonthermal origin of spikes.
Evident energy-dependent time lags are present in a fraction of spikes,
indicative of time-of-flight or Coulomb collision effects. It is also shown that, on
average, spikes lagging in high energy emissions have harder spectra than
spikes exhibiting lags in low-energy emissions.
These numbers are
significantly greater than
Poisson distribution
The negative
occurrence is
nearly ½ of
positive or less
Most of our spikes detected are real signals.
S1
25-100keV
With spike
S2
Without spike
Time cadence 125 ms
Define a spike:
I r(t )  I (t )  I slow (t )  nsig
nsig  3,4,5
wsmt  4,8s
Spike/flare number variations with
different criteria