D. Lamb, C. Graziani, Y. Shirasaki, J-L. Atteia, C. Barraud, K. Hurley,
G. Crew, N. Kawai, G. Ricker, and S. Woosley on behalf of the HETE
M. Boer, J-F Olive, and J-P Dezalay on behalf of the HETE FREGATE Team;
N. Butler, J. Doty, A. Dullighan, G. Prigozhin, R. Vanderspek, J.
Villasenor, T. Cline, J. G. Jernigan, A. Levine, F. Martel, E. Morgan,
G. Monnelly, G. Azzibrouck, J. Braga, R. Manchanda, and G. Pizzichini,
on behalf of the HETE Operations and HETE Optical-SXC Teams;
T. Tamagawa, M. Suzuki, T. Donaghy, M. Matsuoka, K. Torii, T. Sakamoto,
A. Yoshida, E. Fenimore, M. Galassi, Y. Nakagawa, R. Satoh, Y. Urata,
T. Yamazaki and Y. Yamamoto, on behalf of the HETE WXM Team;
At 09:52:03.04 UTC (35523.04 s UT) on 13 Aug 2003, the HETE-2 FREGATE
and WXM instruments detected event H2802, an SGR burst. The WXM
flight localization was correct and was reported in a GCN Notice
at 10:00:17 UT, 8 minutes after the beginning of the burst. The WXM
flight localization of the burst was a circle centered at
WXM Flight: RA = +18h 08m 07s, Dec = -20d 38' 20" (J2000) ,
with a 14 arcminute error radius, which was marginally inconsistent
with a burst from SGR 1806-20 and hence the Notice contained no flag
stating that the source matches a known X-ray source.
Ground analysis of the WXM data produced a localization that was
reported in a GCN Notice at 11:55:09 UT. The WXM ground localization
SNR was 6 in the X-detector and 4.5 in the Y-detector. The
localization can be expressed as a 90% confidence circle that is 7.15
arcminutes in radius and is centered at:
WXM Ground: RA = +18h 08m 11s, Dec = -20d 39' 21" (J2000),
which was reported in a GCN Notice and flagged as a known X-ray source.
Further analysis of the WXM data has produced a refined localization
that can be expressed as a 90% confidence circle that is 6.9 arcminutes
in radius and is centered at
WXM Refined Ground: RA = +18h 08m 11.2s, Dec = -20d 38' 49" (J2000).
Both WXM ground localizations are inconsistent with the burst coming
from SGR 1806-20, suggesting that the burst came from a new source.
The spectrum of the burst can be well-described by a sum of two
blackbody spectra with the following parameters:
n_H = 6 x 10^22 cm^-2,
kT_1 = 4.20 +/- 0.3 keV, and
kT_2 = 10.82 +/- 0.8 keV.
The spectrum of the burst is thus typical of SGR bursts; it is too
hard to be an XRB and too soft to be a GRB.
The burst duration t_90 of the event is 0.16 s in the 7-30 keV energy
band, and 0.14 s in the 30-400 keV band. The fluences of the burst are
7.1 x 10^-7 erg cm^-2, 5.7 x 10^-7 erg cm^-2, and 4.7 x 10^-7 erg cm^-2
in the 7-30, 25-100, and 30-400 keV energy bands, respectively.
We initially waited to report the refined ground location of the burst
in the expectation that confirmation of the location of this possible
new SGR would come from the IPN. However, the Ulysses data for this
event has unfortunately been lost, due to a malfunction of the DSN
We have subjected the WXM localization of the burst to the following
(1) We have localized the burst, using only the WXM wires that were
illuminated by the burst and selecting a background interval that does
not include any peak from the quasi-periodic brightness oscillations of
GRS 1915+105, which was in the FOV of the WXM at the time of the
burst. We find that the center of the localization error circle
changes by < 1 arcminute.
(2) We have localized the burst, using the one WXM X-detector wire that
was not illuminated by GRS 1915+105, the Y-detector wire for which the
relative contamination by GRS 1915+105 was the least, and a 3 s
background interval ending 0.6 s before the beginning of the burst. We
find exactly the same localization for the burst as before, except that
the statistical error is 9 arcminutes instead of 7 arcminutes.
We conclude from these results that the localization of the burst is
not significantly affected by the presence of GRS 1915+105 in the FOV
of the WXM at the time of the burst.
(3) We have localized the burst, assuming that it has a very soft (XRB)
spectrum. We find that the center of the localization error circle
changes by 2.9 arcminutes in the direction directly away from SGR
1806-20. We conclude from this result that assuming a spectrum that
is softer than the spectrum of a GRB increases by a small amount the
angular offset of the burst from the known location of SGR 1806-20.
(4) There is no evidence for saturation of the WXM. However, we have
localized the burst, using only photons in the first 80 ms after the
trigger, when the photon count rate of the burst in the WXM was low.
We find that the radius of the localization error circle increases to
13.6 arcminutes but that the center of the error circle changes by <
2.5 arcminutes. We conclude from this result that the localization of
the burst is not significantly affected by saturation effects.
(5) We have carefully inspected the aspect solution from the optical
cameras, and have concluded that the solution is accurate to better
than 1 arcminute.
(6) We have also localized GRS 1915+105, which was in the field of view
of the WXM at the time of the burst, using two successive peaks in its
quasi-periodic brightness oscillations -- one 25 s before and one 25 s
after the burst. The resulting localization has an error radius of
10.4 arcminutes and is centered 2 arcminutes from the known location of
We conclude from these two results that the aspect of the spacecraft is
well known at the time of the burst, and that other X-ray sources in
the the FOV of the WXM do not significantly affect the localization of
GRS 1915+105, and therefore presumably do not significantly affect the
localization of the burst.
We have therefore been unable to find any reason to doubt the accuracy
of the WXM localization of the burst.
If the probability distribution of the localization were a circularly
symmetric Gaussian, the 15.5 arcminute angular offset of SGR 1806-20
from the center of the WXM localization circle of the burst would
correspond to a chi^2 of 23.2 for 2 DOF. The known location of SGR
1806-20 would then be inconsistent with the localization of the burst
at a formal significance level of 9 x 10^-6. However, although the
real probability distribution of the localization of the burst is
relatively circularly symmetric, it is very non-Gaussian, especially
far out in the tail. The true significance therefore cannot be stated
with any confidence, but it is clearly very small. This strongly
suggests that the burst came from a new source.
We have examined existing ASCA, ROSAT, and Chandra observations of this
field. There are no ASCA sources in the WXM error circle for the
burst. There are several weak ROSAT sources in the error circle for
the burst, but it is not clear that any of them are associated with the
burst source. The FOV of the most recent Chandra observation of SGR
1806-20 did not include the WXM error circle for the burst.
HETE-2 observations of the field are continuing. Other observations
of this possible new SGR are encouraged.