The Astrophysical Journal Letters, 721:L24–L27, 2010 September 20 doi:10.1088/2041-8205/721/1/L24
C
2010. The American Astronomical Society. All rights reserved. Printed in the U.S.A.
FIRST VERY LOW FREQUENCY DETECTION OF SHORT REPEATED BURSTS
FROM MAGNETAR SGR J1550−5418
Y. T. Tanaka
1
, Jean-Pierre Raulin
2
, Fernando C. P. Bertoni
2
, P. R. Fagundes
3
, J. Chau
4
,N.J.Schuch
5
, M. Hayakawa
6
,
Y. Hobara
6
, T. Terasawa
7
, and T. Takahashi
1,8
1
Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, Japan; tanaka@astro.isas.jaxa.jp
2
CRAAM, Presbyterian Mackenzie University, S
˜
ao Paulo, Brazil
3
University of Vale do Paraiba, S
˜
ao Jos
´
e dos Campos, Brazil
4
Radio Observatorio de Jicamarca, Instituto Geof
´
ısico del Per
´
u, Lima, Per
´
u
5
INPE’s Southern Regional Space Research Center, Santa Maria, Brazil
6
University of Electro-Communications, Japan
7
Institute for Cosmic Ray Research, University of Tokyo, Japan
8
University of Tokyo, Japan
Received 2010 June 21; accepted 2010 July 27; published 2010 August 27
ABSTRACT
We report on the first detection of ionospheric disturbances caused by short repeated gamma-ray bursts from the
magnetar SGR J1550−5418. Very low frequency (VLF) radio wave data obtained in South America clearly show
sudden amplitude and phase changes at the corresponding times of eight soft gamma-ray repeater bursts. Maximum
amplitude and phase changes of the VLF signals appear to be correlated with the gamma-ray fluence. On the other
hand, VLF recovery timescales do not show any significant correlation with the fluence, possibly suggesting that the
bursts’ spectra are not similar to each other. In summary, Earth’s ionosphere can be used as a very large gamma-ray
detector and the VLF observations provide us with a new method to monitor high-energy astrophysical phenomena
without interruption such as Earth occultation.
Key words: stars: individual (SGR J1550−5418) – stars: neutron
Online-only material: color figures
1. INTRODUCTION
Very low frequency (VLF; 3–30 kHz) radio waves are re-
flected at Earth’s lower ionosphere and ground, and propa-
gate within the Earth-ionosphere waveguide (e.g., Wait 1962).
Since amplitude and phase of VLF radio waves are sensitive to
the condition of the lower ionosphere, they have been utilized to
investigate the physics of the lower ionosphere. Energetic elec-
trons can precipitate into the ionosphere due to wave–particle
interaction in the magnetosphere and cause VLF signal am-
plitude and phase perturbations (e.g., Kikuchi & Evans 1983).
Soft X-rays from solar flares are also another source of iono-
spheric disturbances, which are detected using VLF signals (e.g.,
Todoroki et al. 2007; Raulin et al. 2010).
Besides these solar–terrestrial events, the lower ionosphere is
also affected by high-energy photons (X-rays and gamma rays)
from extra-solar sources. An ionospheric disturbance caused by
a cosmic gamma-ray burst was first reported by Fishman & Inan
(1988). It suggested that gamma rays deposit their energies in
the lower ionosphere, abnormally ionize the neutral atmosphere
there, and modify the electron density height profile. In addition,
it is known that giant flares from soft gamma-ray repeaters
(SGRs, also called magnetars) significantly affect the lower
ionosphere (Inan et al. 1999, 2007; Tanaka et al. 2008).
Magnetars emit a lot of short-duration gamma-ray flares
repeatedly during active phases (e.g., Woods & Thompson
2006). Typical duration and flux of the short repeated bursts
are 0.1–1 s and 10
−6
erg s
−1
cm
−2
, respectively. Furthermore,
magnetars rarely emit exceptionally bright gamma-ray flares
(giant flares). So far, only three giant flares were recorded. The
first one was detected in 1979, from the source SGR 0526−66
(Mazets et al. 1979). The second and third ones were emitted by
SGR 1900+14 and SGR 1806−20, and they were observed by
satellites in 1998 (Hurley et al. 1999; Mazets et al. 1999; Tanaka
et al. 2007) and 2004 (Terasawa et al. 2005; Hurley et al. 2005;
Palmer et al. 2005; Frederiks et al. 2007), respectively. Since
the fluences were much larger than those of GOES X-class solar
flares by a few orders of magnitude, ionospheric disturbances
caused by these giant flares were detected as sudden and large
amplitude changes of VLF radio waves (Inan et al. 1999, 2007;
Tanaka et al. 2008).
On the other hand, VLF amplitude and phase changes
caused by short repeated gamma-ray bursts from a magnetar
have not been detected so far because of the lack of high
sensitivity of VLF observing systems. On 2009 January 22,
one of the known magnetars SGR J1550−5418 emitted a lot of
short-duration gamma-ray bursts repeatedly (Mereghetti et al.
2009). In this Letter, we report on the first VLF detection
of short repeated gamma-ray flares from this object. VLF
data were provided by the South America VLF Network
(SAVNET) tracking system (Raulin et al. 2009). In Section 2,
we describe details of the SAVNET observations. Comparison
of VLF amplitude and phase changes with gamma-ray fluences
measured by INTEGRAL satellite are presented and discussed in
Sections 3 and 4, respectively. We summarize this Letter in
Section 5.
2. OBSERVATIONS
The VLF data shown in this Letter were obtained by SAVNET,
which was recently installed in Brazil, Peru, and Argentina (see
Raulin et al. (2009) for the details of the SAVNET instrumental
facility). Figure 1 shows locations of a relevant observing station
ATI (Atibaia, S
˜
ao Paulo, Brazil) as well as five VLF transmitters
(NPM, NLK, NDK, NAU, and NAA), and VLF signals from
them have been continuously recorded with the time resolution
L24
No. 1, 2010 FIRST VLF DETECTION OF MAGNETAR SHORT BURSTS L25
Subflare point
NAU
NAA
NPM
NLK
NDK
ATI
Figure 1. VLF propagation path from NPM transmitter (Hawaii) to ATI
observing station (S
˜
ao Paulo, Brazil). Also shown are the locations of other four
VLF transmitters (NLK, NDK, NAA, and NAU). Shaded hemisphere indicates
the nightside part of the Earth at 6:48 UT, when the largest burst occurred (see
Table 1). The part of the Earth illuminated by gamma rays at 6:48 UT is also
drawn by dashed area.
of 1 s. The propagation path from NPM (21.4 kHz) to ATI
is also drawn in Figure 1. Shaded hemisphere in Figure 1
exhibits the nightside part of the Earth at 6:48 UT, when the
most intense gamma-ray flare occurred. The point on the Earth
directly beneath the flare (subflare point) was located at 54.
◦
3S,
14.
◦
0 E, and its position is shown using a cross. The part of the
Earth illuminated by gamma rays at 6:48 UT is illustrated by
dashed area.
Figure 2 shows NPM-ATI amplitude and phase data recorded
from 4:00 UT to 10:00 UT on 2009 January 22. In Figure 3,we
also display an extended view of NPM-ATI data together with
the gamma-ray light curve
9
observed by INTEGRAL satellite
around 6:48 UT (Mereghetti et al. 2009). Due to the high
sensitivity of the SAVNET facility, we can clearly see rapid
amplitude and phase changes at the corresponding times of the
short repeated bursts from SGR J1550−5418. Therefore, we
can robustly claim that the rapid changes were caused by the
short gamma-ray bursts from the magnetar. We listed in Table 1
the properties of the SGR short bursts detected by the NPM-
ATI VLF propagation path. Although these magnetar bursts
were detected at other SAVNET receiving stations, like PAL
(Palmas, TO, Brazil), SMS (S
˜
ao Martinho da Serra, RS, Brazil),
PIU (Piura, Peru), and EACF (Antarctica), in this short Letter
we concentrate on the records from ATI receiving station. A
detailed comparison of VLF phases and amplitudes observed
by the other receivers is out of the scope of this Letter and will
be reported in a subsequent forthcoming article.
3. ANALYSIS
To investigate the influence of gamma-ray irradiation in the
lower ionosphere, we need to characterize the VLF amplitude
and phase data. To do this, we estimated the maximum ampli-
tude (ΔA) and phase variations (Δφ)aswellastherecovery
timescales of the amplitude as follows. First, we extracted the
9
http://www.isdc.unige.ch/integral/ibas/cgi-bin/ibas_acs_web.cgi
4:00
6:00
8:00
10:00
-700
-650
-600
-550
-500
36
32
28
24
Amplitude [dB]Phase [deg]
-0.4
-0.3
-0.2
-0.1
0.0
0.1
2.0
1.0
0.0
02:551:5
-4
-3
-2
-1
0
30
20
10
0
6:40 6:45 6:50
-2.0
-1.0
0.0
10
8
6
4
2
0
8:15 8:20 8:25
Amplitude [dB]
Phase [deg]
UT on 2009 Jan. 22
Figure 2. Amplitude and phase variations of a VLF signal from NPM transmitter
(21.4 kHz), which were observed at ATI (see Figure 1) from 4:00 UT to
10:00 UT on 2009 January 22. Lower figures are background-subtracted blown-
ups at time ranges during which short repeated SGR bursts were detected (see
also Table 1).
(A color version of this figure is available in the online journal.)
100000
80000
60000
40000
20000
0
6:47:30 6:47:45 6:48:00 6:48:15 6:48:30 6:48:45 6:49:00
33
32
31
30
29
28
-670
-660
-650
-640
Phase [deg]
Amptitude [dB]
Count rate [cts/0.05 s]
UT on 22 Januar
y
, 2009
(c) Integral
SPI-ACS
(b) NPM-ATI
(a) NPM-ATI
Figure 3. (a) Blown-up VLF amplitude data from the NPM-ATI path around
6:48 UT. The vertical dashed line shows the time 6:47:57.1 UT, when a relatively
large gamma-ray flare was observed by INTEGRAL (see also Table 1). (b) Same
as (a), but for VLF phase data. (c) INTEGRAL/SPI-ACS light curve around
6:48 UT. Note that the peak of the brightest burst at 6:48:04.3 UT was probably
higher than shown here, due to a saturation problem for high count rates
(Mereghetti et al. 2009).
(A color version of this figure is available in the online journal.)
data before and after each burst, and chose a proper functional
form to represent the baseline level. Most of the baseline levels
can be well fitted by first-order polynomial functions. When the
baseline levels showed curvature, we used second-order poly-
nomial functions to represent them. After subtracting the trends
from the amplitude and phase data, we obtained ΔA and Δφ,
which are tabulated in Table 1. To estimate the typical errors of
ΔA and Δφ, we made histograms of the residuals, which were
distributed around 0 with a Gaussian-like form. Therefore, we
fitted the histogram using a Gaussian and took the variance as a
typical error.
To quantify the recovery timescales of the VLF amplitude
data, we have used the function
f (t) = (Baseline level)
−
F
0
{exp((t
0
− t )/t
fall
)+exp((t − t
0
)/t
rcv
)}
, (1)
L26 TANAKA ET AL. Vol. 721
Tab le 1
Short Repeated Bursts from SGR J1550−5418 Detected by NPM-ATI VLF Data
Start UT INTEGRAL Duration
a
Gamma-ray Fluence
a,b
Amplitude Phase Recovery Timescale Lowering of
on Jan 22
a
ID
a
(s) 10
−5
(erg cm
−2
) Change (dB) Change (deg) of Amplitude (s) Reflection Height (km)
51751.7 85 0.45 1.91 ± 0.01 −0.33 ± 0.03 2.5 ± 0.37.1 ± 1.28.0 ± 1.0
51839.5 93 1.00 1.44 ± 0.01 −0.32 ± 0.03 2.0 ± 0.32.0 ± 0.76.4 ± 1.0
6 41 02.1 108 1.00 2.35 ± 0.01 −0.59 ± 0.04 2.2 ± 0.36.3 ± 0.64.5 ± 0.6
6 44 36.4 117 1.75 1.
93 ± 0.01 −0.91 ± 0.05 2.8 ± 0.35.4 ± 0.65.8 ± 0.6
6 45 13.9 121 1.45 >4.59 ± 0.01 −1.8 ± 0.05 6.5 ± 0.35.9 ± 0.313± 0.6
6 47 57.1 141 0.35 1.82 ± 0.01 −0.74 ± 0.04 2.2 ± 0.5 ...
c
4.5 ± 1.0
6 48 04.3 149 8.15 >27.76 ± 0.03 −4.4 ± 0.04 29 ± 0.59.7 ± 0.260± 1
8 17 29.4 176 6.20 6.59 ± 0.02 −2.4 ± 0.19.3 ± 0.312.0 ± 0.414± 0.5
Notes.
a
Taken from Mereghetti et al. (2009).
b
The energy range is 25 keV to 2 MeV.
c
Unable to determine by fitting.
5
4
3
2
1
0
110
1
10
110
14
12
10
8
6
4
2
0
110
Gamma-ray fluence [10 erg cm ]
-5
-2
(25 keV - 2 MeV)
Gamma-ray fluence [10 erg cm ]
-5
-2
(25 keV - 2 MeV)
Gamma-ray fluence [10 erg cm ]
-5
-2
(25 keV - 2 MeV)
Amplitude change [dB]
Phase change [deg]
Recovery timescale [s]
)c( )b( )a(
Figure 4. (a) Relation between observed amplitude changes and gamma-ray
fluences (25 keV to 2 MeV) measured by INTEGRAL satellite (Mereghetti et al.
2009). Values are tabulated in Table 1. (b) Same as (a), but for observed phase
changes. (c) Same as (a), but for recovery timescales estimated by fitting.
where F
0
is a typical amplitude decrease, t
fall
is a falling time,
and t
rcv
is a recovery timescale. We fitted the data using this
function and determined t
rcv
for each burst (see Table 1).
4. DISCUSSION
4.1. Amplitude and Phase Changes
In Figures 4(a) and (b), we plot ΔA and Δφ against the gamma-
ray fluence from 25 keV to 2 MeV for each burst (see also
Table 1). Although there are not many data points, possible
correlations between ΔA and Δφ, and gamma-ray fluences are
seen. We note that similar correlations were also reported in the
case of X-ray solar flares (McRae & Thomson 2004; Pacini &
Raulin 2006).
We can understand these evidences of correlation in terms
of the lowering of the reflection height due to gamma-ray
ionization. Under a typical undisturbed nighttime condition,
VLF waves are thought to be reflected at ∼85 km (e.g., Carpenter
et al. 1997). When gamma rays are directed onto the Earth, they
deposit most of their energy in the lower ionosphere, ionize
the neutral atmosphere there, and produce free electrons. The
typical altitude where these free electrons are produced depends
on the photon energy. For example, by using Monte Carlo
simulation Inan et al. (1999) reported that 3 keV and 10 keV
photons mainly ionize the atmosphere at ∼82 km and 60 km,
respectively. Similar calculations have shown that gamma-ray
illumination increases electron number density below ∼85 km,
depending on the photon spectrum (e.g., Brown 1973;Baird
1974; Tanaka et al. 2008). Consequently, the VLF radio waves
are reflected at a lower altitude than usual, and hence the phase
of propagating VLF waves is advanced.
Due to the lack of observation, the exact photon spectrum for
each burst was not reported so far. But following Mereghetti
et al. (2009), we assume the spectral shape of an optically thin
thermal bremsstrahlung with kT = 40 keV. Then, the number of
higher-energy photons increases as the fluence goes up. Higher-
energy photons can penetrate deeper at low altitude and increase
the electron number density there. As a result, the reflection
height becomes lower as the gamma-ray fluence increases.
By treating the propagation of VLF radio waves using the
mode theory (Wait & Spies 1964), we estimated the reduction
of the reflection height ΔH from the phase change Δφ.In
the following, we have assumed that the lower ionosphere is
isotropic and a sharply bounded medium (Wait & Spies 1964),
and we have used a phase velocity expression given by Wait
(1959). Then, the relation between Δφ and ΔH can be expressed
as (Inan & Carpenter 1987
)
Δφ
dΔH
−
2πf
hc
h
2R
e
+ C
2
n
C
n
=
(
2n − 1
)
λ
4h
, (2)
where d is the length of the disturbed region along the great circle
path, f is the wave frequency, h is the typical reflection height,
c is the speed of light, R
e
is Earth’s radius, λ is the wavelength
of the VLF radio wave, and n is the order of the waveguide
mode. For a long propagation distance and a normal nighttime
reflection height of h ∼ 85 km, Wait & Spies (1964)showed
that the second mode (n = 2) would be dominant. Therefore,
we were able to calculate ΔH from the observed Δφ for each
burst and these values are tabulated in Table 1. We note that
ΔH calculated by using Equation (2) is a rough estimate, and
Monte Carlo simulations are required to obtain more accurate
values. Nonetheless, this gross estimation would be meaningful
to qualitatively consider the effect of gamma-ray illumination.
Next, we consider a mechanism for the observed decrease
of VLF wave amplitude during the gamma-ray illumination.
We can understand it on the basis of the altitude dependence
of the collision frequency ν
e
between electrons and neutral
atoms. ν
e
is higher for lower altitudes and is often modeled as
ν
e
= 1.816×10
11
exp(−0.15z), where z is the altitude measured
in km (e.g., Wait & Spies 1964). Consequently, as the reflection
height becomes lower, VLF radio waves are more attenuated,
and hence their amplitudes decrease.
These hints of correlation suggest that it would be possible to
deduce a gamma-ray fluence from ΔA and Δφ. We note that these
relations are applicable only for this particular VLF frequency
(21.4 kHz). There are also another uncertainties which might
affect ΔA and Δφ such as the altitude profile of the ambient
electron number density. Nonetheless, we claim that Earth’s
No. 1, 2010 FIRST VLF DETECTION OF MAGNETAR SHORT BURSTS L27
ionosphere can be used as a new gamma-ray “detector” and VLF
data can provide a unique information on incident gamma-ray
fluences, even if satellites in space were not able to observe it.
Therefore, we stress here that this VLF method is a new potential
technique for monitoring high-energy transient phenomena in
the universe, once we know in advance which source is active.
4.2. Recovery Timescale
We plot in Figure 4(c) the recovery timescale of each burst
against the gamma-ray fluence. We did not find any significant
correlation between the two quantities, and all the recovery
timescales were in the range of 2–12 s. Since the t
rcv
are
longer than the burst durations (see Table 1), the observed VLF
amplitude and phase time profiles are different from the gamma-
ray light curves. We also fitted the phase data with a similar
function of Equation (1) and calculated the t
rcv
. Again, we did
not find any significant correlation between gamma-ray fluences
and recovery timescales.
As shown in Figure 2, recovery time profiles can be well rep-
resented by an exponential-like function of a single parameter
t
rcv
. On the other hand, in the case of magnetar giant flares, two
different recovery timescales were reported (Inan et al. 2007;
Tanaka et al. 2008). Namely, an initial rapid recovery of a few
seconds was followed by a long enduring recovery lasting for
>1 hr. The long-duration recovery is interpreted as due to the
neutralization of positive and negative ions at an altitude be-
low 60 km (Inan et al. 2007), which means that ionization by
gamma rays mainly occurred at such a low altitude. In fact,
it is known that the photon spectrum and gamma-ray fluence
of giant flares are much harder and higher than those of short
repeated bursts (Woods & Thompson 2006). Lack of such a
long-duration recovery in our VLF data suggests that the spec-
tra of short repeated bursts were relatively soft compared to that
of giant flares.
As shown above, we interpreted the VLF amplitude and phase
changes as due to the lowering of the reflection height. In this
case, faster recovery timescales are expected for larger gamma-
ray fluences, because the electron attachment rate is a negative
function of altitude (Rowe et al. 1974). However, as shown
in Figure 4(c), we did not find such a trend. This might be
due to different spectrum for each burst, contrary to what we
have assumed in this Letter following Mereghetti et al. (2009).
The harder the spectrum, the lower the reflection height and the
faster the recovery time. Monte Carlo simulations are required
to confirm this possibility, and the results will be reported in a
subsequent article.
5. SUMMARY
We have detected, for the first time, ionospheric disturbances
caused by short repeated gamma-ray bursts from a magnetar.
Amplitude and phase changes of VLF propagating waves are
correlated with gamma-ray fluences. This can be understood in
terms of the lowering of the reflection height. While satellites
in space cannot continuously observe the whole sky due to
Earth occultation, Earth’s ionosphere can monitor it without
interruption. VLF observations provide us with a new method
to monitor high-energy transient phenomena of astrophysical
importance.
Y.T.T. is supported by a JSPS Research Fellowship for
Young Scientists. J.P.R. thanks FAPESP (Proc. 06/02979-0),
CNPq (Proc. 304433/2004-7), and MACKPESQUISA funding
agencies.
REFERENCES
Baird, G. A. 1974, J. Atmos. Terr. Phys., 36, 1565
Brown, R. T. 1973, Nature, 246, 83
Carpenter, D. L., et al. 1997, J. Geophys. Res., 102, 347
Fishman, G. J., & Inan, U. S. 1988, Nature, 331, 418
Frederiks, D. D., Golenetskii, S. V., Palshin, V. D., Aptekar, R. L., Ilyinskii, V.
N., Oleinik, F. P., Mazets, E. P., & Cline, T. L. 2007, Astron. Lett., 33, 1
Hurley, K., et al. 1999, Nature, 397, 41
Hurley, K., et al. 2005, Nature, 434, 1098
Inan, U. S., & Carpenter, D. L. 1987, J. Geophys. Res., 92, 3293
Inan, U. S., Lehtinen, N. G., Lev-Tov, S. J., Johnson, M. P., Bell, T. F., & Hurley,
K. 1999, Geophys. Res. Lett., 26, 3357
Inan, U. S., Lehtinen, N. G., Moore, R. C., Hurley, K., Boggs, S., Smith, D. M.,
& Fishman, G. J. 2007, Geophys. Res. Lett., 34, 8103
Kikuchi, T., & Evans, D. S. 1983, J. Geophys. Res., 88, 871
Mazets, E. P., Cline, T. L., Aptekar’, R. L., Butterworth, P. S., Frederiks, D. D.,
Golenetskii, S. V., Il’Inskii, V. N., & Pal’Shin, V. D. 1999, Astron. Lett., 25,
635
Mazets, E. P., Golentskii, S. V., Ilinskii, V. N., Aptekar, R. L., & Guryan, I. A.
1979, Nature, 282, 587
McRae, W. M., & Thomson, N. R. 2004, J. Atmos. Sol.-Terr. Phys., 66, 77
Mereghetti, S., et al. 2009, ApJ, 696, L74
Pacini, A. A., & Raulin, J. P. 2006, J. Geophys. Res. (Space Phys.), 111, 9301
Palmer, D. M., et al. 2005, Nature, 434, 1107
Raulin, J. P., Correia de Matos David, P., Hadano, R., Saraiva, A. C. V., Correia,
E., & Kaufmann, P. 2009, Earth Moon Planets, 104, 247
Raulin, J. P., et al. 2010, J. Geophys. Res., 115, A07301
Rowe, J. N., Mitra, A. P., Ferraro, A. J., & Lee, H. S. 1974, J. Atmos. Terr.
Phys., 36, 755
Tanaka, Y. T., Terasawa, T., Kawai, N., Yoshida, A., Yoshikawa, I., Saito, Y.,
Takashima, T., & Mukai, T. 2007, ApJ, 665, L55
Tanaka, Y. T., Terasawa, T., Yoshida, M., Horie, T., & Hayakawa, M. 2008, J.
Geophys. Res. (Space Phys.), 113, A07307
Terasawa, T., et al. 2005, Nature, 434, 1110
Todoroki, Y., Maekawa, S., Yamauchi, T., Horie, T., & Hayakawa, M. 2007,
Geophys. Res. Lett., 34, 3103
Wait, J. R. 1959, Proc. IRE, 47, 998
Wait, J. R. 1962, Electromagnetic Waves in Stratified Media (New York:
Elsevier)
Wait, J. R., & Spies, K. P. 1964, NBS Tech., 300
Woods, P. M., & Thompson, C. 2006, in Soft Gamma Repeaters and Anomalous
X-ray Pulsars: Magnetar Candidates, ed. W. H. G. Lewin & M. van der Klis
(New York: Cambridge Univ. Press), 547
