The following is a paper published at SPIE Conference (19-23 July 1999, Denver, CO; VOl 3768, pp 444). While this web page for the GCN/NEAR operation is not in the standard form as the other GCN operations (i.e. I am using this SPIE paper), it does provide most of the information (albeit in a non-standard format).
The GCN-NEAR are not real Notices (yet) like the other operations. Only the lightcurves are posted to the GCN web page NEAR Light Curves. Someday the Lightcurves will be available as distributed Notices (like the rest of the GCN system).
TABLE OF CONTENTS:
S.D. Barthelmy(a,b), T.L. Cline(a), P. Butterworth(a,e), D. Palmer(a,b), J. Trombka(a), T. McClanahan(a), R. Gold(c), K. Hurley(d)
(a) NASA-GSFC, Greenbelt, MD, 20771
(b) Universities Space Research Association
(c) Applied Physics Laboratory
(d) UC Berkeley, Berkeley, CA
The on-board flight software for the Near Earth Asteroid Rendezvous (NEAR) spacecraft was modified to produce continuous 1-sec sampled rate information from the shield of the X-ray and Gamma Ray Spectrometer (XGRS) instrument. Since the XGRS shield can also detect Gamma Ray Bursts (GRB), this rate information can be used in combination with the GRB detections by the Ulysses and near-Earth GRB instruments as part of the Interplanetary Network (IPN) to triangulate the source direction of GRBs. It is the long baseline of NEAR (~1.5 AU) combined with the Ulysses baseline (~5 AU) that makes small error box locations possible. We have developed an automated system to analyze the periodic telemetry dumps from the NEAR spacecraft. It extracts this new data type, scans the rate information for increases which are plausibly of GRB origin, and combines these with the GRB detections from the other spacecraft. Because the processing is automated (for all spacecraft), the time delay to produce the triangulated positions is kept to a minimum, up to 48 hours (all but a few minutes of which is due to the inherent delay of the TM dumps). This automated processing and distribution of the GRB locations is done within the GRB Coordinates Network (GCN) system. About 60 localizations per year with errors ranging from a few to tens of arcminutes are expected. These rapid precise localizations may provide about 10 times the rate currently provided by the WFC and NFI instruments on BeppoSAX.
Keywords: Gamma Ray Burst, GRB, Automated Analysis & Distribution
The origin of gamma ray bursts (GRBs) remained a well-known mystery in astrophysics for nearly a quarter of a century until a soft x-ray afterglow was found with the BeppoSAX mission1. Unlike the intense and brief emission in gamma rays, this radiation persisted long enough to refine the source location with greater accuracy, and even to realign the spacecraft to improve on the source location, permitting the follow-up study that found the first optical2. Spectroscopic analyses of faint but long-lasting optical afterglows in turn have, in a number of cases, found red shifts3-4 that clearly demonstrate a cosmological origin of GRBs and that GRBs are the most powerful explosions in Nature after the Big Bang itself. Time is of the essence in these follow-up observation efforts, since the fading of the afterglows, in the radio as well as optical regimes, makes it difficult for astronomers to unambiguously identify the source.
A remarkable new discovery earlier this year that provides a new twist to this story is that of the prompt optical burst that accompanied a GRB on 1999 Jan 23rd5. This discovery was possible only because of the use of the GRB Coordinate Network (GCN), which was devised to distribute GRB source locations in real time - as the gamma ray burst is being monitored, and because of the construction of a robotic telescope that was built to take advantage of the GCN. Follow-up efforts have evolved over the last decade since Compton-GRO was launched with delays decreasing from months, to weeks, to days, and to hours with BeppoSAX, and finally with the Akerlof event, to seconds. It is now apparent that the optical flash can begin at the time of the gamma ray emission, not only hours later as appeared to be suggested by the earlier studies, and that time is even more critical a parameter than ever in the race to maximize the full description of a gamma ray burst and its wide-band characteristics. The interplanetary networks that have been in use to determine precise source locations of GRBs for over two decades have had a chaotic history of intermittent successes, due to launch failures and other unfortunate circumstances. The addition of the NEAR mission to the IPN now completes a full, three-way, long-baseline network for the first time since the deaths of Pioneer Venus Orbiter and Mars Observer in the early 1990s, and the first such network since the advent of the GCN with its automated, computerized global data distribution6. We describe the incorporation of NEAR into the IPN using the GCN in this paper.
2. THE NEAR MISSION
The NEAR mission (launched Feb. 1996) is the first of the Discovery series spacecraft7. Its instrument complement is designed to study near-earth asteroids -- Mathilde and Eros in particular. Figure 1 shows the basic configuration of the NEAR spacecraft. The original schedule called for a velocity-decreasing burn in Dec. 1998 placing the spacecraft into an orbit around Eros starting Dec. 1999. But that burn aborted prematurely, and the new schedule calls for the Eros orbital phase of the mission to begin Jan. 2000 and last for about 13 months. The orbital phase is important to the current GRB effort because of (1) the increase of XGRS instrument on-time up to nearly 100%, and (2) the increase of the telemetry downlinks from 3 days per week (M, W, & F) to 3-4 times per day, 7 days a week.
Figure 1. Schematic representation of the NEAR spacecraft.
3. THE NEAR-XGRS INSTRUMENT
The X-ray Gamma-Ray Spectrometer (XGRS) instrument is composed of two completely separate detectors8. Three gas proportional counters operating in the 1-10 keV range to make the X-Ray Spectrometer (XRS). The Gamma-Ray Spectrometer (GRS) has a central NaI detector surrounded by a Bismuth Germanate (BGO) anti-coincidence shield. The combined instrument is XGRS. It is this BGO shield that we are using to detect the GRB gamma-rays. It operates in the 0.3- 10 MeV range. See Figure 2 for the details of the GRS instrument and BGO shield configuration. Because of its mounting location and configuration on the spacecraft, the BGO shield is sensitive to GRBs from approximately 2_ steradians. The original science goals for the GRS are to measure basic elemental abundances in Eros and to provide a coarse mapping (roughly octants) of those abundances. The count rate in the GRS BGO shield is continuously sampled every second and this rate information is included in the telemetry stream during all periods when the GRS instrument is on. See section 4 for the details of the GRS shield data stream.
4. MODIFICATIONS TO THE XGRS INSTRUMENT
Two post-launch modifications to the XGRS flight software were implemented to make our GRB observations possible. The first was to include in the normal telemetry stream a continuously 1-sec sampling sequence of rate data from the XGRS BGO shield. These data are in packets with type number 15. Each packet contains 166 contiguous 1-sec rate samples plus a time stamp. This modification was uploaded and activated in 1997. These data are present in the telemetry stream during phases of the mission when the high telemetry downlink rate is available. During some portions of the cruise phase of the mission and during the orbit phase, the downlink telemetry rate plus the fraction allocated for GRB monitoring will be considerably reduced. There is insufficient capacity to transmit the continuous shield rate data. To retain the ability of obtaining the shield count rate data during bursts for analysis, an on-board trigger algorithm was implemented. This on-board trigger continuously monitors the shield rate data looking for an ~5 sigma increase and captures a 164-sec window of data around this trigger time (with a pre-trigger capture of 32 sec). Then during a downlink session, only this window of shield rate data is transmitted with the appropriate trigger time reference. This modification was uploaded and activated on May 24, 1999.
TBD Figure 2. Schematic cross-section of the GRS instrument.
5. THE NEAR ON-BOARD TIME
Since the IPN timing triangulation method (see section 7) is dependent on knowing the absolute time when a burst is detected by each spacecraft, it is critical to have an accurate absolute time reference on-board each spacecraft contributing to the IPN solution. Using periodic DSN ranging measurements, the NEAR on-board system clock was found to be drifting by about 20 msec/day. This was independently confirmed by direct measurements of the very intense soft gamma repeater (SGR) giant burst of 1998 August 27. It was detected by all the instruments participating in the IPN, including NEAR9, with timing accuracy as good as 2 milliseconds. The results clearly showed that the NEAR clock had drifted (as expected) by over 2 seconds, and that although the situation of continued drift could be calibrated with repeated SGR occurrences, the IPN utility would benefit with the on-board reestablishment of a correct clock.
The on-board clock fix, made in the Spring of 1999, has been checked using the GRB of 1999 May 10 plus its optical transient location that permitted the determination of the time the GRB would be expected at the NEAR spacecraft. That time has been shown to agree with the measured time to within 0.1 second. Even though the GRB time history on NEAR is monitored with only 1.0-second-wide binning, by phasing the GRB count-rate profiles from the GGS-Wind-Konus experiment a least-squares fit demonstrated an accurate fit to the time required to match the detection profile of that event at NEAR. Future such checks will be possible both with future SGRs and with future GRBs that have associated optical transients. Note that 0.1-second timing accuracy translates into better than 0.5-arcminute directional accuracy with a baseline of 1.5 AU. In general, the fluctuations in the GRB count-rate profiles contribute greater errors, resulting in the final source locations of several arcminute precision.
6. THE AUTOMATED DATA PROCESSING
The flow of data from the spacecraft to our ground processing system is a complicated sequence of actions and time delays. The basic sequence is as follows: Telemetry downlinks for NEAR are scheduled for every Monday, Wednesday, and Friday in the early afternoon (EST) (during the cruise phase). The data from the spacecraft (including XGRS) from the previous 48 to 96 hours is played back from the on-board solid-state data recorders. Each downlink session lasts about 8 hours, and there is no real-time downlink simultaneous channel. After the end of a session, it takes 10-30 minutes to process the data to a point where it is usable. This processing involves (1) resequencing the data into true chronological order, (2) correcting bit errors from the transmission process, (3) removing duplicate data sets (sometimes more than one ground station is receiving and recording the downlink data or sometimes portions of the data from the solid-state recorder are retransmitted), (4) merging engineering, housekeeping and auxiliary support data for use by the scientific analysis teams, (5) stripping out of the XGRS instrument data into a separate file, (6) ftp-ing the data from the JPL facility to JHU-APL, (7) ftp-ing the data from APL to GSFC, and finally (8) processing the data at GSFC looking for GRBs. The total of the on-board storage delays, transmission delays, and ground processing delays yields a range of 8.5 to 96 hours from when a potential GRB occurred to when the burst light curve is available to be included in the IPN processing.
During the orbit-phase of the mission (when NEAR is orbiting around the Eros asteroid, starting Feb. 2000), the telemetry downlink sessions will be increased to 3-4 times per day, 7 days per week. There will also be a real-time channel being transmitted during the playback downlinks. This will decrease the total time delay to 0.5 to ~10 hours, with an average of ~6 hours. This decrease in the total time delay from the time of the burst will greatly improve the scientific utility of the IPN solutions.
Steps 1 through 6 are standard for most deep-space missions. Steps 7 and 8 require more explanation. The process at GSFC happens as follows: (1) a daemon automatically spawns an ftp process every 20 minutes to check on the availability of XGRS data files at the APL facility. (2) If there are any new files, they are automatically copied to GSFC. (3) Each new file is checked to see if it is a completely new data set or if it is an augmented data set previously transferred. The later is possible because previously downlinked data sometimes have gaps due to telemetry problems and so retransmissions are commanded for the following session. (4) The type 15 GRS shield rata data packets are extracted from the entire XGRS data stream and assembled in proper chronological order. (5) This sequence of 1-sec sampled shield rate data is scanned for increases in the count rate (see section 6.1 for a description of the triggering algorithm). (6) When a rate increase is found, 80 sec of pre-trigger plus 80 sec of post-trigger rates are written to a file. (7) This light curve file of a putative burst is e-mailed to the automated IPN processing system in Berkeley, CA (see section 7 for a description of the IPN processing). And finally, (8) when the IPN system combines this XGRS-detected burst light curve with two or more light curves from other spacecraft, then the IPN solution is e-mailed back to the GCN system (see section 8) for distribution to the burst follow-up community.
6.1. Trigger Search Algorithm
On the ground, the continuously sampled rate data from the XGRS shield are scanned for increases in the count rate. These increases can be due to non-astrophysical (i.e. non-GRB) causes, but when the XGRS putative-burst light curve is combined with other burst light curves from other spacecraft, the non-GRB increases are eliminated. The trigger scanning algorithm goes as follows: (1) a background counting rate is found in a sliding window of width 64 sec plus the standard deviation for that background rate is calculated (sigma). (2) Then each rate sample 32 sec after the sliding background window is tested to see if it is greater than N sigma above the corresponding background (N is typically 6-8). (3) A second rate increase test of any 2 samples within the next 10 sec window greater than M sigma (typically M is 4). The N and M values are adjusted to yield about 1 trigger per 24-hour period of data. On average, about half of these are false, non-GRB related triggers, and the remaining half are due to real GRBs (a few/week).
6.2. Secondary Effort
In addition to looking for GRBs directly in the XGRS shield rate data, the auto-NEAR system receives times from the GCN system of when GRBs were detected by other spacecraft (e.g. Wind-Konus). The program then uses those times to initiate a more sensitive search (i.e. lower effective threshold) for the GRB within the XGRS data. Because of the various distances of the spacecraft from the NEAR spacecraft and the direction of the burst, there is a window within which the burst wave-front would have crossed the NEAR spacecraft. Therefore, a window of plus or minus the light travel time for those distance variations must be searched centered around the burst time.
7. THE INTERPLANETARY NETWORK
Once the GRS light curves are extracted, they are e-mailed to a system maintained by K.Hurley (PI of the Ulysses GRB instrument, UC Berkeley). There they are scanned for the starting time. The starting time plus the distance of the Ulysses spacecraft from NEAR determine the window in time the Ulysses data must be scanned for a potential matching light curve. Since Ulysses has periodic telemetry dumps (like NEAR), it is possible the Ulysses data has not arrived. If it has not been received from the Ulysses mission operations, then the search is put on a pending queue.
The Interplanetary Network (IPN) uses the differences in the burst detection times by widely separated spacecraft to determine the direction of the burst10. The arrival time difference between only two spacecraft yields an annulus on the sky. The center of that annulus is the R.A.,Dec. direction of the line connecting the two spacecraft, and the radius of the annulus is the arccos(c*delta_T/D), where c is the speed of light, delta_T is the arrival time difference, and D is the distance between the two spacecraft. In practice an improved value for delta_T is determined by finding the maximum cross-correlation of the two light curves for varying relative time delays. The uncertainty in this process is a function of the fluence of the burst, the amount of structure in the burst light curves (the size of the binning of the count rate data), the accuracy of the on-board clocks, and the accuracy of the knowledge of the spacecraft locations in the solar system. The typical width of the annulus is under several arcmin for a Ulysses-NEAR location. When three spacecraft detect a given burst, then a second (non- redundant) annulus can be calculated. Typically, this annulus crosses the first annulus in two places, yielding two small error boxes for the location of the burst. The degeneracy in the two locations is removed by using the rough location from CGRO- BATSE (typically, an error circle of 3-5 deg in diameter). If only NEAR-XGRS and CGRO-BATSE detect the burst, then the resulting annulus is combined with the BATSE location circle to yield a long thin segment of the annulus (typically, several arcmin wide by several degrees long). The rates of occurrence for these two location scenarios is several per month for the 3-spacecraft error boxes and several per week for the annular segments.
8. THE GRB COORDINATES NETWORK
The GRB Coordinates Network (GCN) is a system of custom equipment, computers, and software to calculate, collect, and distribute locations of GRBs in realtime in a few seconds. The GCN system started out in 1993 as the BACODINE system11 which processed the real-time telemetry from Compton-GRO, in particular the Burst and Transient Source Experiment (BATSE). Custom electronics captures and decommutates the telemetry stream extracting the BATSE data. A series of computers and programs is used to automatically unpack the BATSE data, scan for GRBs, calculate appropriate backgrounds for subtraction purposes, calculate the direction of the burst, and then distribute that location to a set of clients around the world. In May 1997, the BACODINE system was renamed to the GCN system after it started collecting and distributing GRB locations detected by other instruments and spacecraft. The current list of contributors of burst information is: CGRO-COMPTEL, RXTE-PCA/-ASM, Wind-Konus, BeppoSAX-WFC/-NFI, and the IPN; plus location information on extreme-UV transients from the ALEXIS spacecraft and x-ray transients (novae, etc.) from RXTE-ASM and -PCA. In the future, GCN will distribute locations from the HETE-II (launch in Jan. 2000) and INTEGRAL (launch 2001). GCN clients are thus provided with a single location and a single protocol for obtaining all available cosmic gamma-ray burst and transient information. The GCN system is always looking for new sources of astrophysical transient information that it can collect and distribute to its ever growing list of client instruments to make follow-up observations. These instruments are already in place and many are fully automated (with no humans in the loop), and so are optimally suited for making rapid follow-up observation of short-lived astrophysical transient source objects.
The GCN system was designed to get and distribute the information with the minimum amount of time delay, as such several of the distribution methods have time delays in the range of 0.3 sec to 10 sec, typically. The methods are shown in Table 1. The phone-modem method is the fastest, but suffers from significant toll charges. The Internet socket method is essentially as fast (0.3 2.0 sec for >95% of the packets) and is essentially free for any operation with Internet access already in existence. For both the phone and socket methods, pre-arranged formatted packets of information are exchanged between the GCN system and the site. These packets contain the burst (R.A.,Dec.) location and a measure of the uncertainty in the location, the burst date and time, the burst intensity, the spacecraft-instrument that detected the burst, and some miscellaneous supporting information that allows the user to assess the quality and importance of the burst and formulate a follow-up observing plan. Examples of the software needed to implement the Internet socket method have been developed and are available to the client site for their convenience. The phone and socket methods are most suitable for site instruments that are fully automated and fast-slewing allowing them to make the most rapid follow-up observations. The e-mail method is more suited for human recipients. A token-value formatted message is generated and sent to the designated list of recipients. For clients who are more mobile not at a fixed instrument location or who want immediate notification regardless of their access to their normal e-mail account, they can receive abbreviated information on alpha-numeric pagers and cell-phones. For the vendors that offer such services, an e-mail message is sent to their pin_number account where it is on-forwarded to their pager or cell-phone. The GCN system also maintains a web site which contains all the location notice information from the various spacecraft-instrument contributors plus additional information like burst light curves and reports from follow-up observations made (the GCN Circular service). The web site pages are updated within 15 sec of the specific notice being distributed. It is an example of "pull technology", whereas the other distribution methods are "push technology". More information about the GCN system can be found at the web site12: the methods used and a description of what types of instruments and operations are appropriate for GCN notices and how to arrange to receive GCN burst location notices (by any of the methods).
The GCN system allows each site to custom tailor the type of information they receive and the method used. Filters can be chosen based on day vs. night status at the site (for optical sites), visibility above the site¹s local horizon, time-of-day and declination band filters, error box size filters, plus filtering on the spacecraft-instrument source.
Table 1. GCN Distribution Methods and Typical Distribution Times
Distribution Method Typical Delivery Times Comments [sec] Phone-Modem 0.3 Continuous phone/modem connection Internet Socket 0.1 - 2.0 Fast & suited for automated instruments E-mail 5 - 30 To any network address Pager & Cell-phone 60 - 180 Ra,Dec,Time,Intensity displayed on pager or phone. Web site 15 Time is time to update the page
The project would like to express its appreciation for the skilled and dedicated work provided by Teresa Sheets (GSFC) who has written much of the software for auto-NEAR, and to Doug Holland (APL) who provided much of the data flow and telemetry format information for NEAR-XGRS.
1. E. Costa, F. Frontera, J. Heise, M. Feroci, J. In't Zand, F. Fiore, M.N. Cinti, D. Dal Fiume, L. Nicastro, M. Orlandini, E. Palazzi, M. Rapisarda, G. Zavattini, R. Jager, A. Parmar, A. Owens, S. Molendi, G. Cusumano, M.C. Maccarone, S. Giarrusso, A. Coletta, L.A. Antonelli, P. Giommi, J.M. Muller, L. Piro, R.C. Butler; ³Discovery of an X-ray afterglow associated with the gamma-ray burst of 28 February 1997², Nature, 387, 783, 1997.
2. J. van Paradijs, P.J. Groot, T. Galama, C. Kouveliotou, R.G. Strom, J. Telting, R.G.M. Rutten, G.J. Fishman, C.A. Meegan, M. Pettini, N. Tanvir, J. Bloom, H. Pedersen, H.U. Nordgaard-Nielsen, M. Linden-Vornle, J. Melnick, G. Van Der Steene, M. Bremer, R. Naber, J. Heise, J. In't Zand, E. Costa, M. Feroci, L. Piro, F. Frontera, G. Zavattini, L. Nicastro, E. Palazzi, K. Bennet, L. Hanlon, A. Parmar; ³Transient optical emission from the error box of the gamma-ray burst of 28 February 1997²; Nature, 386, 686, 1997.
3. M.R. Metzger, S.G. Djorgovski, S.R. Kulkarni, C.C. Steidel, K.L. Adelberger, D.A. Frail, E. Costa, F. Frontera,; ³Spectral constraints on the redshift of the optical counterpart to the gamma-ray burst of 8 May 1997.², Nature, 387, 879, 1997.
4. G. Djorgovski, S.R. Kulkarni, R. Goodrich, D.A. Frail, and J.S. Bloom; ³GRB980703: Spectrum of the proposed optical counterpart², GCN Circ. 137, 1998.
5. C. Akerlof, R. Balsano, S. Barthelmy, J. Bloch, P. Butterworth, D. Casperson, T. Cline, S. Fletcher, F. Frontera, F.; G. Gisler, J. Heise, J. Hills, R. Kehoe, B. Lee, S. Marshall, T. McKay, R. Miller, L. Piro, W. Preidhorsky, J. Szymanski, J. Wren, ³Observation of contemporaneous optical radiation from a gamma-ray burst², Nature, 388, 400, 1999.
6. T. L. Cline, S. Barthelmy, P. Butterworth, F. Marshall, T. McClanahan, D. Palmer, J. Trombka, K. Hurley, R. Gold, R. Aptekar, D. Frederiks, S. Golenetskii, V. Il'Inskii, E. Mazets, G. Fishman, C. Kouveliotou, and C. Meegan; "Precise GRB source locations from the renewed interplanetary network"; Astronomy and Astrophysics (in press), 1999.
7. C.T Russell, editor; ³The Near Earth Asteroid Rendezvous Mission², page3, Klewar, 1997.
8. C.T Russell, editor; ³The Near Earth Asteroid Rendezvous Mission², page169, Klewar, 1997.
9. T.L. Cline, E.P. Mazets, S.V. Golenetskii, ³SGR 1900+14², 1998, IAU Circ. 7002.
10. K. Hurley, M.S. Briggs, R.M. Kippen, C. Kouveliotou, C. Meegan, G. Fishman, T.L. Cline, M. Boer; ³The Ulysses Supplement to the BATSE 3B Catalog of Cosmic Gamma-Ray Bursts², ApJS, 408, 120, 1999.
11. S.D. Barthelmy, T.L. Cline, N. Gehrels, J.R. Kuyper, G.J. Fishman, C. Kouveliotou, C.A. Meegan; ³BACODINE, The Realt-Time BATSE Gamma-Ray Burst Coordinates Distribution Network², Second Huntsville GRB Workshop, AIP Proceedings, 307, 1994.
12. S.D. Barthelmy, GCN Web site URL: http://gcn.gsfc.nasa.gov/gcn/, 1999.