HETE Burst Alerts
One of the key features of the HETE satellite is its ability to
calculate precise localizations of GRBs on board within seconds of
burst onset, and then to transmit the burst localizations to
the ground as soon as they have been calculated.
The HETE-2 satellite utilizes a low-rate VHF transmitter to
continuously broadcast the burst information;
on the ground, an array of listen-only
burst alert stations (BAS)
receive the data and transmit them to the MIT Control Center.
Once received at MIT, burst information are immediately
relayed to the
GRB Coordinate Distribution Network (GCN)
at the Goddard Space Flight Center for distribution to interested
On this page, we describe
A recent modification to the ground burst location selection processes
is described in the imaging quality section .
Burst information downlinked by HETE
Information about a GRB will come to the ground in two ways:
- The results of real-time analyses performed on the spacecraft
are transmitted via the VHF to the burst alert stations.
These results will be of moderate quality, as the spacecraft
processors are computationally limited.
- The raw data taken by all science instruments are transmitted
via the S-band data link. With these data, sophisticated
ground analyses can refine the burst position by factors
Burst analysis procedures
The analysis of HETE burst data occurs in several different stages
on several different fronts:
- Real-time analyses are performed on the spacecraft,
resulting in a quick but unrefined calculation of the
burst localization and spectrum. The results of these
analyses are distributed over the VHF link within seconds
of the burst detection.
- After the downlink of the raw data, automated analysis
engines run through the GRB data and come up with a
refined position. These results will be available 15 minutes
to two hours after the GRB, depending on where in the
spacecraft's orbit the burst occurred.
- Once a human analyst has been notified and reaches a
computer, a manual analysis of the data will be performed.
We expect that the results of such an analysis will
yield a significant improvement over the previous two
methods. The results of this analysis will be available
several hours after the burst.
What HETE sends to the GCN...
...in real time
The current criteria for the real-time distribution of
GCN Notices are as follows:
- If the burst is detected using photons in the 5-80 keV or
30-400 keV bands, a GCN Notice of type S/C_Alert
is sent out regardless whether a position has been
determined or not.
- If a position is determined on board and it is considered
significant enough, the RA and declination of the burst
will be distributed in a type S/C_Update GCN
Each additional position determined on board (each
with higher significance than all determined before)
will result in a new Notice of type S/C_Update .
- Once the on-board processing of data near the time of the
trigger is complete, there will be no more immediate results
from the spacecraft, so a summary Notice of type
S/C_Last is distributed. The S/C_Last
Notice has the best answers that the flight software can
This method of distribution of GCN Notices results in a few common
- If there is no significant position calculated in real time
on board, there will be no burst coordinates in any GCN
Notice. If ground analysis reveals a position, it will
be sent out as a type HETE_Gnd_Analysis Notice.
- Because the Burst Alert Station coverage is not always 100%,
there can be gaps in the reception of burst data from the
spacecraft. If the flight analysis of a burst is over before a
Burst Alert Station is seen, the full analysis of the burst will
be sent in two Notices, one of type S/C_Alert and the
other of type S/C_Last . This means that a Notice of
type S/C_Alert could, in principle, contain the
coordinates of the burst.
...and after Ground Analysis
Ground analysis of a burst begins as soon as the full burst data reach
MIT after a Primary Ground Station contact (from a few minutes to over
an hour after the burst, depending on where in its orbit HETE was at
the time of the burst).
Automated software performs standard analyses of the downlinked data,
and a human is notified to make the final decisions.
A followup GCN Notice, of type HETE_Gnd_Analysis , will be
distributed under the following circumstances:
- There was no position calculated on board, but ground analyses
reveal a significant position.
- There was a position calculated on board and ground analyses
can improve the coordinates and/or reduce the error box size.
- There was a position calculated on board, but there is actually
no significant position in the data.
In general, if there is a position in a real-time GCN Notice, it
should be considered accurate.
If there is no position in a real-time GCN Notice, a position may
be forthcoming within an hour or so of the original Notice.
If a burst position was distributed and it is wrong because of software
or operator error, a HETE_Gnd_Analysis message will be distributed;
if no position was distributed, no HETE_Gnd_Analysis message
will be sent.
HETE-2 Burst Detection
The HETE mission has three instruments on board which can generate
WXM, and the
Fregate data are examined in two broad energy channels: 5-80 keV and
30-400 keV. The WXM data are from 2-30 keV, and the SXC data are from
1.5 to 12 keV. There are three processors, running in parallel,
examining these data for signatures of a burst.
- The Fregate data are searched for counts excesses on four
different timescales: 20 ms, 160 ms, 1.3 s, and 5.2 s.
The threshold for a count excess to be considered significant
is different for each timescale, but it is generally around
Such an excess must be seen in two of the four Fregate
detectors on the same timescale for a burst to be considered
real. This analysis is done on a dedicated on-board DSP.
- The WXM data are examined on multiple timescales between
80 ms and 10 s. The thresholds vary from timescale
to timescale, but all are near 5 sigma. This analysis is
done on one of the on-board transputers.
- Fregate data are also analyzed on multiple timescales in
a manner identical to that for the WXM data on the same
transputer the WXM data are analyzed by.
- SXC data can be used to create a continuous series of
cross-correlation maps using a
dedicated DSP, and a burst is registered when the peak
of the cross-correlation map exceeds a threshold.
Because of the difficulties with the SXC hardware, SXC
triggering is currently not operating.
When a burst is detected, the real-time spacecraft notification
- the energy range of the burst (1.5-12 keV, 2-30 keV, 5-80 keV,
or 30-400 keV)
- the timescale of the trigger (from 20 ms to over 10 s)
- the S/N or the peak count rate of the burst
HETE-2 Burst Imaging
Once a burst has been detected on board, all other flight instrument
computers are notified: the WXM and SXC begin their attempts
to image the X-rays from the burst, and the optical system, which
has immediate knowledge of the spacecraft attitude and aspect,
delivers the spacecraft orientation to the the mission operations
center at MIT via the real-time VHF link and the
Both the WXM and SXC search data from seconds before the burst
trigger to minutes after, looking for the image of the burst.
The WXM software matches the shadow pattern on the detector with
template patterns, looking for a best fit; the SXC looks for
peaks in the cross-correlation map. In either case, if a
significant position is found in either instrument, its location
is sent to the ground in real time.
Once the positions and their significances are received on the
ground, the RA and declination of the image are calculated and,
if the significances are high enough, transmitted to the GCN.
At present, SXC positions are not sent to the GCN automatically,
but rather only after ground analysis.
Once a burst position has been determined on board and sent to
the ground, the on-board software continues to look for better
images using shadow patterns collected during different time
intervals. If a new burst position is calculated which has
a higher signficance than the preceding one(s) , that position
will be sent to the ground via the VHF.
Determining the quality of HETE real-time positions
After extensive analysis of GRBs and XRBs detected by HETE, we have
determined a method of verifying with >90% certainty that a
real-time position is correct. This method involves measurement of
the "lightcurve signal-to-noise" and the "image signal-to-noise"
of a burst in each of the X and Y modules of the WXM. (Real-time
burst positions as calculated by the SXC flight software are
currently not distributed to the GCN).
- The lightcurve S/N describes the signal-to-noise of
the excess over background in the WXM lightcurve (LC) in each
of the two modules (X & Y). The WXM lightcurve is obtained by
summing the data over all position bins at 320ms time resolution.
The flight software chooses background and burst time intervals,
and performs a standard background subtraction in each detector.
The resulting net signals are divided by their respective
root-variances, to yield the quantities reported here.
Note that the lightcurve S/N values are not directly related to
trigger S/N, since the trigger operates on 80ms resolution data,
and may in fact come from FREGATE data rather than from WXM data.
The light-curve S/N is designed measure the overall signal-to-noise
in the actual data used to obtain the location,
so as to complement the image S/N.
- The image S/N describes the peak signal-to-noise in the
cross-correlation of the background-subtracted image with the mask
in each of the two modules (X & Y). The background-subtracted
signal across each WXM detector is cross-correlated with the mask,
and the resulting cross-correlation functions are mean-subtracted
and divided by their respective root-variances, yielding SNR as a
function of source projection angle in each detector. The quantities
reported are the peak values of those functions.
We have determined that bursts that are well-localized by the WXM have
image S/N and lightcurve S/N >3.0. We have also found that those
which do not have image and lightcurve S/N >3.0 are moderately well
imaged by the flight software if the total image S/N (the quadrature sum
of the X and Y image S/N) is >3.7 and the measured incident angle
of the burst is <30 degrees in both X and Y modules.
As a result, HETE positions distributed in real time from the spacecraft
are in one of two categories:
- Category I: The image and lightcurve S/N all exceed 3.0,
so the position is distributed with a 90% error radius of 12-14
- Category II: Not all of the image and lightcurve S/N exceeds
3.0, but the quadrature sum of the image S/N from the WXM X and Y
detectors is > 3.7, and neither the X nor the Y incident angle
exceeds 30 degrees, so the position is distributed with a 90% error
radius of 30 arcminutes.
We are confident that, with this scheme in place, the burst coordinates
as calculated by the flight software will be correct >90% of the time.
However, the HETE GCN interface has been modified recently (4/2002) to
accommodate those observers who would like to make their own estimate
of the quality of a real-time burst localization. Now included in the
GCN message are
- The image S/N from the X and Y modules of the WXM
- The lightcurve S/N from the X and Y modules of the WXM
- The longitude of the HETE spacecraft at the time of the trigger.
The higher the image and lightcurve S/N, the more reliable the
localization will be. Low values of image and lightcurve S/N are
typically associated with events localized at the edge of the
The only exceptions to this rule has been
seen during particle events, where both the image and lightcurve
S/N values can been extremely high. These events were quite frequent
during solar maximum, but have become less so since the beginning of
2002. To determine whether a trigger was caused by a particle event,
Since this solar maximum cycle is now on the decline, particle events
are going to be very rare, so each real-time location distributed by
HETE should be considered valid at the level of significance measured
by the error circle radius.
- Check the spacecraft longitude at the time of the trigger: if
the event occurred at 280 degrees E longitude +/- 40 degrees,
it could be a particle event.
- Check the
the planetary Kp index monitor at NOAA: if the Kp value is
4 or higher, there is a geomagnetic storm in progress and the
probability of a particle event is high. If Kp is under 3, the
event is not likely to be due to particles.