Transcript ppt

Timing analysis of unusual
GRB 090709A observed by
Suzaku Wide-band All-sky
Monitor
Wataru.Iwakiri (Saitama university)
Tuneyoshi Kamae (SLAC/KIPAC), Masanori Ohno (ISAS), Yukikatsu
Terada, Makoto.S.Tashiro (Saitama Univ.), Yujin.E.Nakagawa (RIKEN),
Atsumasa.Yoshida, Kazutaka Yamaoka (Aoyama gakuin Univ.) , Kazuo
Makishima (RIKEN & Univ. of Tokyo) and Suzaku-WAM team
GRB090709A
Sec
BAT(10 – 350 keV)
High- amplitude
periodicity (about 8 sec)
●T0 : 07:38:34 (UT) ●T90 : 89 sec
●Fluence (15-150 keV): 2.57(±0.03) x10-5 erg/cm2
●The 8 sec periodicity reported by Markwardt et al (GCN 9645).
We will perform the statistical significance of the periodicity.
2
The Light curves
WAM TRN data
1 s resolution
WAM: BGO scintillator
(70 – 5000 keV)
WAM BST data
128 ms resolution
Konus data
128 ms resolution
Konus: NaI scintillator
(50 – 220 keV)
BAT: CdZnTe semiconductor
(15 – 350 keV)
BAT data
128 ms
resolution
3
The Light curves
WAM TRN data
1 s resolution
・Only using the BAT data, the
periodicity is less than 3 sigma.
(A.De Luca et al.2010; Cenko et al 2010)
・These light curves contain of
GRB signal and instrumental
white noise (Poisson noise).
・ The GRB signal is common in
the 3 detectors, but instrumental
noise is individual.
・To improve the signal-tonoise ratio,
we sum up the 3 data.
Problem:
The time resolution and
data length is different.
WAM BST data
128 ms resolution
Konus data
128 ms resolution
BAT data
128 ms
resolution
4
Emulate the WAM light curve
・For 1 sec bin WAM-TRN data, we re-distributed photon into
8 sec times bin according to Poisson statistics and covers
whole duration with WAM-BST data.
BST data exist
only 64 s.
Sec
TRN data (1 s
time resolution)
5
Emulate the WAM light curve
Sec
・For 1 sec bin WAM-TRN data, we re-distributed photon into
8 sec times bin according to Poisson statistics and covers
whole duration with WAM-BST data.
Emulate data
(128 ms time
resolution)
BST data exist
only 64 s.
6
Emulate the Konus light curve
Sec
・Konus covers only 100 sec data, we extended the light curve
with a Poisson noise with the average of bgd level.
7
Emulate the Konus light curve
Sec
・Konus covers only 100 sec data, we extended the light curve
with a Poisson noise with the average of bgd level.
8
0.9
0
Cross
correlation
WAM v.s
Konus
・ The cross correlation functions show
that good coincidence of these data.
BAT v.s Konus
0
0.9
0
Sec
WAM(70 – 5000 keV)
BAT(15 – 350 keV)
Konus(50 – 220 keV)
WAM v.s
BAT
0.9
Sum up the 3 data
9
0.9
0
Cross
correlation
WAM v.s
Konus
・ The cross correlation functions show
that good coincidence of these data.
BAT v.s Konus
0
0.9
0
Sec
WAM(50 – 5000 keV)
Summed up
BAT(15 – 300 keV)
the 3 data
Konus(10 – 770 keV)
WAM v.s
BAT
0.9
Sum up the 3 data
10
The Power spectrum of GRB 090709A
8 sec
11
What is a Color noise ?
a noise : ∝f-α
Light curve
Power spectrum
White noise
Alpha=0
Pink noise
Alpha=1
Red noise
Alpha=2
12
Color noise index of GRBs
Suzuki et al.
2002
The distribution of the color noise index alpha
for 297 BATSE burst.
13
The Power spectrum of GRB 090709A
White noise
Color noise
14
Estimation of white noise
bgd light curve
Power spectrum
The white noise level is
derived the bgd power
spectrum from
– 100 s to –20 s before trigger.
15
Estimation of the color noise index
Power-law index: 1.9
Suzuki et al.
2002
16
Comparison the simulation and observation
Power
Green:
average of 1000 times
simulated data
Magenta:
99.85 % level of deviation
Red: 99.9% level
Null hypothesis
probability is about 0.1 %
Green:deviation of
1000 times
simulated data
Blue:deviation of
observed data
Period [sec]
17
Discussion
• A period of 8 sec is comparable to typical time
scale of soft gamma-ray repeater in our galaxy.
• However, we see no evidence of periodic
oscillation in the afterglow (A.De Luca et al 2010) but
see features we often see from GRB afterglows.
(See Nakagawa et al. Poster 064 in this conference;
A.De Luca et al 2010; Cenko et al 2010) .
• The origin of the periodicity is a GRB inner
engine ??
18
Summary
• GRB090709A detected by WAM, BAT, and
Konus.
• We summed up 3 detector’s data to obtain
good signal-to-noise ratio data by emulation
of the light curve.
• Comparison the 1000 times color noise
simulation and observed power spectrum
density, we detected the 8 sec periodic signal
with the null hypothesis probability of 0.1 %.
19