This page is an addendum to the GCN Circular 6155 (D. Denisenko, O. Terekhov) published on 2007 Feb. 27 and the extended version of the paper submitted to Astronomy Letters (Pis'ma v Astronomicheskiy Zhurnal). It contains the color images and animations which cannot be inserted to the printed version of journal, but are essential for illustrating the discovery.
As a result of the systematic search for the archival plates taken at the Palomar and Siding Spring observatories in the course of all sky survey, several cosmic gamma-ray bursts were selected whose error boxes were covered (entirely or in part) within a day after the event. Inside the error circle of bright long GRB 920925C an optical transient of magnitude 17.8 is found on the plate taken 6 hours after the burst. The position of an object falls inside the IPN annulus. Analysis of the event properties allows to claim the transient to be the optical afterglow of the cosmic gamma-ray burst which has occured almost 4.5 years before GRB 970228.
Comment: This page contains the data related to GRB 920925C. In order not to overfill one html-document with the images, the article is divided in three parts. You can find the detailed discovery story, list of the events studied and the findings on the other burst at the separate pages:
For almost 10 years GRB 970228 was considered to be the first gamma-ray burst with the detected optical afterglow (V=21.3 measured 21 hr after the trigger, see Groot et al., IAUC 6584). More than a hundred optical afterglows have been observed since then, the brightest one being GRB 030329 (z=0.1685) with R=12.6 at T0+1.5hr and V=14.5 at T0+6hr. Mostly the long bursts accompanied by X-ray emission are observed in visual and infrared bands. Optical afterglows are usually fading with time as t-0.8-t-1.2 during the first day(s).
There were three gamma-ray bursts detected on Sep. 25th, 1992 by WATCH instrument (Sazonov et al., 1998), all within 2.5 hours. The brightest and the longest of them is the third one. The table below lists all three bursts with their trigger time (UT), duration, fluences in 10-7 erg/cm2 in two energy bands and their ratio. The last column indicates the confirmation of bursts by the instruments on other spacecrafts (Ph - PHEBUS/Granat, B - BATSE/CGRO, U - Ulysses).
-------------------------------------------------------- Burst Trigger Duration Fluence Hardn Other name h m s T90, s 8-20 keV 20-60 ratio S/C -------------------------------------------------------- W920925 20 30 42 57.10 22.3 44.5 2.00 PhU W920925b 21 45 20 12.98 10.9 20.3 1.87 BU W920925c 22 46 24 282.17 120.3 207.9 1.73 U --------------------------------------------------------
Let us consider GRB 920925C in more detail.
GRB 920925C is among the longest and brightest bursts observed by WATCH instrument during its operation onboard the Granat observatory. It has the third longest duration out of 87 events with measured parameters, the third largest fluence in the soft energy band and the 10th one at hard energies.
The light curve of GRB 920925C in two energy bands is given on the figure (top panel showing the counts per second in 8-20 keV range, bottom panel - 20-60 keV). Temporal resolution is 7 sec. Time is given in seconds since 22:46:24 UT.
The burst was localized by WATCH instrument to the circle centered at:
R.A. = 330.80, Dec. = 25.48 deg (22h03m12s, +25o28'48", Epoch J2000.0)
with 3-sigma error radius 0.34 deg, or 0.39 deg including systematic error (Sazonov et al., 1998).
WATCH error box falls inside the field 532 of the 2nd Palomar Observatory Sky Survey (POSS-II). This area was covered by the Red plate SF03598 in October 1990 and IR plate SN06310 in July 1995. POSS-I Red and Blue (O) plates were taken in August 1950, Quick-V plate - in October 1983.
Blue Palomar plate SJ04837 (A1F7, region name XJ532) with 55-min exposure was taken on 1992 Sep. 26 between 04:51-05:46 UT (6.1-7.0 hr after the burst). The WATCH position of GRB 920925C lies 0.31 deg from the plate center. We have downloaded the 15'x15' images from DSS Plate Finder.
Note: if you enter coordinates 330.80, 25.48 and click Find plates you will be informed that Blue plate was taken on 1992-09-25 at 04:85:00. This is wrong! The procedure converting the plate date from YYYY.YYYY HH.HH format to YYYY-MM-DD hh:mm is taking 2000.0 for 2000 Jan 0.5 while the POSS plate dates were packed using 2000.0 = 2000 Jan 1.5. So 1992.7351 corresponds to JD 2448891.5, or 1992-09-26, and exposure start time 04.85 - to 04:51 UT.
3x3 mosaic with 3' overlap between nearby plates covers the entire WATCH error circle. Then images from the 1992 blue plate were blinked against the 1950 blue plate in search for possible optical transient. Since the spectral response of POSS-II blue (J) and POSS-I blue (O) plates are slightly different, the suspect objects were checked on R, IR and Quick-V plates.
On the very first set of 15'x15' images we found a star-like object which is present on POSS-II Blue plate only. The animation of 1950 and 1992 blue plates is shown on the next picture (101"x101" crop at 4x zoom).
The object has S/N ratio of 45 and FWHM=2.66". Using the nearby stars from the USNO-A2.0 catalogue, its blue magnitude was measured to be 17.8+/-0.1. The following J2000.0 coordinates were determined relative to UCAC2 catalogue:
R.A. = 22 03 31.26, Dec. = +25 25 01.4 (330.8802 deg, +25.4171 deg)
This is 0.096 deg (6', or 1 sigma) from the center of WATCH error circle.
Below is the color-combined image of the same 101"x101" field around the OT shown on the animation. IR plate from 1995 was used as a Red channel, R plate from 1990 - as Green and 1992 B plate as Blue. It is obvious that the fainter objects missing on POSS-I Blue plate are real stars or galaxies which were not detected in 1950 due to a shorter exposure (12 min versus 55) and sensitivity of the O plates shifted towards UV band (300-510 nm for POSS-I O compared to 350-580 nm for POSS-II J, see info on surveys).
Note also a slightly elongated object from the IR plate right above the image center. It is located at 1.5" W, 2.8" N from the optical transient and could be a host galaxy of GRB 920925C, but its detection significance is only 2 sigma. Deep follow-up observations are necessary to confirm the presense of this extended object and to measure its redshift if it is indeed related to GRB 920925C.
As noted above, GRB 920925C was also detected by the Ulysses spacecraft. This allowed to localize the burst to the box with an area of 475 square arcmin (Hurley et al., 2000). This is a 3.6 times reduction of WATCH error circle of 1720 square arcmin.
Intersection of the IPN annulus with the WATCH error circle is shown on the next figure. Half-width of annulus (3-sigma) is 0.086 deg, or 5'.
The next image shows the position of the optical transient from the blue plate taken 6.1-7 hours after the burst. It falls inside the IPN annulus within 0.058 deg (2 sigma) from its center.
The optical transient under large magnification looks slightly elongated towards south-west. We have selected several stars of the 15-17 magnitude and compared their images on Blue and IR plates. This elongation is present in many stars:
While the same stars are elongated towards south-east on the infrared plate:
This effect is especially obvious on the animation.
The start of 55-min long exposure for the blue plate corresponds to the hour angle 0:38E. That means the telescope and the mount crossed the meridian during the exposure. On the other hand, the 85-min long IR plate was taken at the 1:28E hour angle. Plate centers and orientations are identical. The object in both cases is located near the center of 7x7 deg plate.
FWHM parameter was measured for 28 stars on the plate ranging in magnitude from 15 to 19. The result is shown at the following graph.
Position of the optical transient is in Pegasus, 34.5 deg away from ecliptic. There were five known asteroids brighter than 18th magnitude inside the 7x7 degree plate. Their coordinates at 5:00 UT on 1992 Sep. 26 and hourly motions produced at Ephemeris Generator for Palomar Mountain - DSS  are listed in the table.
Object number name R.A. (J2000.0) DEC dRA*cosD d(DEC)/dt APmag --------------------------------------------------------------------- 2001 BU41 21 58 59.12 +23 50 49.5 -37.34 11.38 16.89 (24972) 1998 FC116 21 54 32.35 +24 03 20.4 0.60 -63.83 16.69 (71818) 2000 UH19 22 16 12.36 +25 00 27.2 -20.25 -52.43 18.01 (89501) 2001 XR48 22 00 09.08 +22 30 39.0 -17.77 -25.12 18.14 (94695) 2001 XN35 21 59 03.87 +23 38 29.8 -21.09 -21.10 17.72 ---------------------------------------------------------------------
All asteroids have moved by more than 25" during 55 minute exposure. Clicking on the asteroid name you can see their images from the Blue plate. Magnified images of (24972) and (89501) are shown below (2'x2' FOV at 2x zoom).
Note the profiles of asteroid tracks have drops in intensity corresponding to the meridian crossing time. That can be an explanation of star images being elongated towards the south-west. This shift was obviously small, of the order of 1", and is not visible on the bright stars with FWHM over 4". But it is sensible for the fainter stars. It can also serve as an indirect prove that the optical transient was present at the same position during the whole exposure time and was not a satellite glint, for example.
|Asteroid (24972) moved on the sky towards south. On the image it corresponds to the rightward direction (Y increases). Note the gap around Y=415.|
|Asteroid (89501) moved on the sky towards south-west. It corresponds to the rightward direction on the image (increasing X and Y). Note the gap around two thirds of the track. Telescope crossed the meridian 38 minutes after the start of exposure and 17 minutes before the end.|
There should be about 5 transneptunian objects brighter than 18m over the whole sky. On 1992 Sep. 26 Pluto, 2005 FY9 and 2003 EL61 were located 99, 117 and 119 degree from the GRB 920925C, correspondingly. Two or three other bright TNOs can still remain unknown hiding somethere on the sky. Hence the probability of the transient being a yet undiscovered TNO is about 10-6.
There should be about 10000 cataclysmic variables brighter than 18m in the peak over the whole sky (supposing logN-logS distribution with the index 2.5: there is one CV brighter than 8m in outburst and ~100 CVs brighter than 13m). Cataclysmic variables of UGSS and UGSU types spend about 10%-3% of time in outburst, higher amplitude UGWZ stars - about 0.1%. Hence the probability of the transient being a CV in outburst is about 1/300. If the object was indeed a CV in outburst, there must be a star of ~22-23 magnitude in quiescence on its place.
The position of optical transient was covered several times by NEAT project using the same Palomar 48-inch Schmidt camera, but already equipped with CCD. 26 images were downloaded from SkyMorph website. They cover the following dates: 2000 June 09, 2000 Aug. 05, 2000 Sep. 04, 2000 Nov. 22, 2001 Aug. 04, 10, 11 and 16, 2001 Sep. 23 and 2001 Nov. 25. No object is present at the place of 1992 Sep. 25 flash on any of images. Sum of 12 best ones is shown of the next figure.
Three infrared exposures of the field in J, H, K bands were taken in course of 2MASS project on 1999 Nov. 15. Below is the combined image with J as B channel, H as G and K as R. It is not deep enough to tell anything for sure.
The second (brighter) peak of GRB 920925C was also detected by PHEBUS instrument of the GRANAT observatory in 100-1600 keV energy range. The burst did trigger PHEBUS, but was only recorded in two blocks of electronics out of three, and the telemetry data were corrupted. This is why it was not included into PHEBUS catalogue of gamma-ray bursts back in 1995. In February 2007 we have reconstructed the light curve using available PHEBUS spectra of the burst in time-to-spill mode. Unfortunately the background information is incorrect since it was accumulated over one-minute interval between the first and second peaks. Thus the zero level shown by the dotted line on the plot is just an approximation.
Time of PHEBUS trigger marked as 0 on the X axis corresponds to 22:49:03.641 UT. Using this lightcurve it can be possible to improve the IPN annulus. Temporal resolution of PHEBUS lightcurve is better than that of WATCH (approximately 1 sec compared to 7). It should be noted, however, that BGO detectors of PHEBUS operated in a harder energy range than NaI and CsI detectors of Ulysses and WATCH, and the burst is showing remarkable spectral evolution already from 8-20 to 20-60 keV bands.
There was another Blue Palomar plate of area 0532 (UJ04890) taken on Oct. 6, 1992, but with 3 minute exposure only. It is not available in the digitized format. Supposing the 1/t decay of afterglow, the optical transient should have faded to 21.5-22.0 by Oct. 6th. The limiting magnitude of the short exposure plate, on the contrary, is hardly deeper than 18m. Still it can be useful to inspect that plate in order to check for the possible object or put the upper limit on the brightness of the transient.
We conclude that the object found inside GRB 920925C error box on blue Palomar plate taken 6.1-7.0 hours after the burst is most probably the optical afterglow of this GRB. If so, it is the first gamma-ray burst afterglow captured almost 4.5 years before GRB 970228.
- Niels Lund, Gamma-Ray Burst Positions - Are They Reliable? (1995!!!) Contains position for GRB 920925 as 22 02 00, +25 15 (J2000.0) and 30'x120' error box combined from IPN annulus and preliminary WATCH localization.
"We therefore hope that the new data presented here will be useful for burster counterpart searches in different energy ranges as well as for studying possible correlation in GRB positions" - Sergey Sazonov, WATCH catalogue of GRB (1998!!!)
"However, there is still a need for less precise GRB localizations of older bursts, for several reasons. For example, the discovery of bright optical emission coincident with one burst indicates that searches through archival optical data may reveal other examples of this interesting phenomenon." - Kevin Hurley, Ulysses Supplement to WATCH catalogue of GRB (2000!!!)
"However, there's another method of 'hunting' the bursts. Search through photographic archives comes to help. Since the bursts occur at random positions at random moments, amateur astronomers sometimes capture right places at the right times." - Denis Denisenko, Let's search gamma-ray bursts in optics (2003!!!). Text itself is in Russian, but it has a link to the list of well localized bursts since 1990 which occured in the dark time over horizon for European part of Russia. The list contains GRB 920925 at 22:46:24 UT in Pegasus, 15 degrees from M15 globular cluster.
All four documents quoted above were found via Google search after the detection of the possible optical counterpart on 2007 Feb. 10th at 23:30+0300.
This work has intensively used the
Sky Survey produced at the Space Telescope Science Institute based on
photographic data obtained at the Palomar Observatory with the help of
California Institute of Technology and funds from NSF, NGS, Sloan Found.,
Samuel Oschin Found. and Eastman Kodak Corporation.
Thanks to Alexey Tkachenko for the help in resurrecting PHEBUS data analysis software.
Thanks to Aleksey Kuzmin for storing the CD with WATCH data.
Thanks to Alexei Pozanenko for useful discussions after work on the way home in metro at 1 a.m.
Thanks to Minor Planet Mailing List members for opening my eyes on the errors in the DSS plate finder at STScI site.
Discovery was made after watching the movie "The Butterfly Effect". Warning: don't try to repeat it - that may affect the past events!
Feb. 24th, 2007 20:30+0300
Last updated: Feb. 27th, 2007 19:00+0300
Denis Denisenko firstname.lastname@example.org
To my GRB Page | To my Astronomy Page | To my Main Page