nasa logo

National Aeronautics and

Space Administration

Appendix G to the

NASA

RESEARCH ANNOUNCEMENT

for the

COMPTON GAMMA RAY OBSERVATORY
GUEST INVESTIGATOR PROGRAM


TABLE OF CONTENTS

I. INTRODUCTION G-1

II. OSSE GUEST INVESTIGATOR PROGRAM G-2

A. Introduction and Scientific Objectives G-2

B. The OSSE Instrument G-2

C. Instrument Parameters and Capabilities G-8

D. Data Reduction and Analysis G-10

E. OSSE Guest Investigator Opportunities G-18

F. OSSE Observation Definition Form G-20

G. OSSE Data Product Availability G-21

III. COMPTEL GUEST INVESTIGATOR PROGRAM G-22

A. Introduction and Scientific Objectives G-22

B. The COMPTEL Instrument G-23

C. Instrument Parameters and Capabilities G-28

D. Data Reduction and Analysis G-30

E. COMPTEL Guest Investigator Opportunities G-34

F. COMPTEL Facilities for Guest Investigators G-43

IV. EGRET GUEST INVESTIGATOR PROGRAM G-47

A. Introduction and Scientific Objectives G-47

B. The EGRET Instrument G-48

C. Instrument Parameters and Capabilities G-51

D. Data Reduction and Analysis G-51

E. EGRET Guest Investigator Opportunities G-54

F. Feasibility Evaluation for EGRET Observations G-55

G. Specification of EGRET Observations G-57

H. Data Products to be Provided to EGRET Guest Investigators G-60

I. Use of EGRET During Cycle 6 and Beyond G-60

V. BATSE GUEST INVESTIGATOR PROGRAM G-63

A. Introduction and Scientific Objectives G-63

B. The BATSE Instrument G-64

C. Data Reduction and Analysis G-67

D. Data Analysis Environment G-71

E. BATSE Guest Investigator Opportunities G-71

F. BATSE Flight Performance and Observation Sensitivity G-72

G. Defining an Observation G-78

H. Data Products G-78

VI. CGRO SCIENCE SUPPORT CENTER G-83

A. General G-83

B. Guest Investigator Support G-83

C. Data Catalog and Archives G-85

D. Other User Support Activities G-88

THE COMPTON GAMMA RAY OBSERVATORY AS A GUEST INVESTIGATOR FACILITY

I. INTRODUCTION

The Compton Gamma Ray Observatory (Compton), was conceived, designed, and developed as a Principal Investigator class mission but has subsequently become an observatory class facility accessible to the international astrophysics community. Its four distinct instruments are optimized to perform simultaneous observations of specific targets or regions. These instruments combine to cover over five decades of energy with more than a factor of 10 improvement in sensitivity throughout the band, and with improvements in both spectral and spatial resolution in selected energy regions over previously flown instruments. The purpose of this Appendix is to describe Compton's capabilities and the opportunities available to Guest Investigators.

The approach to the Compton Guest Investigator Program takes into account the complicated nature of the data reduction and analysis processes for each of the four instruments. These instruments are described below. Details of the Program are instrument dependent and these are described in separate sections. However, some aspects of the Program are common to all instruments. There are various ways in which Guest Investigators can participate in Compton observations and data analysis. The descriptions listed below are not intended to limit or otherwise constrain the nature of proposed investigations, but only to provide examples of the primary forms of scientific investigations and collaborations presently anticipated.

At this stage of the mission, most Investigators propose to work more or less independently of the Instrument Team, especially if existing data is being used. It may be advisable to contact the Science Support Center (SSC) and/or Instrument Teams should be consulted regarding any technical feasibility issues or policy issues. Alternatively, an Investigator can propose to work directly and closely with one or more of the Compton Instrument Team scientists or CGRO-SSC in the analysis of data received from the Observatory. Guest investigators wishing to be present at the instrument team's prime site (UNH, MSFC, GSFC, or NRL) during the observation of a target assigned to the Guest are welcome to be present and to participate in the data analysis from the start if they so choose. Generally speaking, flux estimates for targets are available in about 7-10 days. Full processing of target data takes 1-4 months.

II. OSSE GUEST INVESTIGATOR PROGRAM

A. Introduction and Scientific Objectives

The Oriented Scintillation Spectrometer Experiment (OSSE) will conduct a broad range of observations in the 0.05-250 MeV energy range. Major emphasis is placed on scientific objectives in the 0.1-10.0 MeV region with a limited capability above 10 MeV, primarily for observations of solar gamma-rays and neutrons and observations of high-energy emission from pulsars.

The scientific objectives of OSSE include observations of the following:

The OSSE instrument has been designed primarily to investigate localized sources and to a lesser extent to map Galactic diffuse emissions with its 3.8 x 11.4 (FWHM) field-of-view. During the normal two week observing period, OSSE will observe two sources using its two-source per orbit observing capability. This will result in approximately 50 source observations per year.

B. The OSSE Instrument

The Oriented Scintillation Spectrometer Experiment (OSSE) has been designed to undertake comprehensive observations of astrophysical sources in the 0.05-10 MeV energy range. Secondary capabilities for gamma-ray and neutron observations above 10 MeV have also been included, principally for solar flare studies. The use of NaI(Tl) scintillation techniques provides large area and high sensitivity in the 0.05-10 MeV region.

The highest sensitivity is maintained by implementation of an observational technique which minimizes susceptibility to background variations due to orbital geomagnetic latitude effects and spallation produced radioactivity. This technique uses an offset pointing capability to modulate a celestial source's contribution to the observed gamma-ray flux on a time scale short compared to the background variations, so that it can be reliably extracted from the background. Normally, OSSE performs two-minute viewing on and off the targeted source. Background variations are modulated by the spacecraft's orbital period of about 90 minutes.

The offset pointing capability also enables the experiment to undertake observations of selected sources (e.g., transient objects) with a significant independence of their location on the sky or on pointing restrictions which may be imposed by the spacecraft. For example, this capability will be used to monitor solar flare activity without impacting the planned observational program of the other Compton experiments. It will also be used to achieve a two-source per orbit capability wherein a secondary source will be studied during orbital periods of earth occultation of the primary source. In addition, this capability is used to allow OSSE to respond to gamma-ray bursts on time scales of a few seconds if they occur near the OSSE scan plane.

Four identical detector systems are used to obtain the operational flexibility to meet the variety of scientific objectives previously outlined. Each of OSSE's four detectors operates, to a large degree, as an independent experiment. They have independent electronic systems and pointing systems. The synchronization of the four detectors is provided by the central electronics, which provides the data acquisition timing and coordination of the pointing directions. As shown in Figure II-1, each of the detectors is mounted in a single axis orientation control system which provides offset pointing over a range of 192 in the spacecraft X-Z plane. The detectors are generally operated in co-axial pairs. While one detector of a pair is observing the source, the other detector can be offset to monitor the background. After a programmable time interval, typically two minutes, the detectors will interchange observation directions by opposite rotations. Each detector's maximum rotation (or slew) rate is 2o per second.

Figure II-2 displays the major components of one of the OSSE detectors. The primary element of each detector system is the NaI(Tl) portion of a 330-mm diameter NaI(Tl)-CsI(Na) phoswich consisting of a 102-mm thick NaI(Tl) crystal optically coupled to a 76-mm thick CsI(Na) crystal. Each phoswich is viewed from the CsI face by seven 89-mm diameter photo-multiplier tubes (PMTs), providing an energy resolution of 8% at 0.661 MeV. Active gain stabilization is used to maintain this energy resolution by individually adjusting the gain of each of the seven PMTs (the absolute gain of the OSSE phoswich remains stable to within 0.1% during a typical two week observing period. Utilizing the differing scintillation decay time constants of NaI(Tl) and CsI(Na), the detector event processing electronics incorporates pulse-shape analysis for the discrimination of events occurring in the NaI crystal from those occurring in the CsI, allowing the CsI portion of the phoswich to act as anticoincidence shielding for the NaI portion. A tungsten alloy passive slat collimator, located directly above the NaI portion of each phoswich, defines the gamma-ray aperture of the phoswich detector, providing a 3.8x 11.4 FWHM rectangular field-of-view throughout the 0.1-10 MeV energy range. Each phoswich aperture is covered by a charged particle detector (CPD) consisting of a 508-mm square by 6-mm thick plastic scintillator which is viewed by four 51-mm diameter photo-multiplier tubes and is used for charged particle background rejection (energy losses greater than a nominal threshold of 0.20 MeV in the CPD trigger rejection of coincident energy losses in the phoswich). Both the phoswich and tungsten collimator are enclosed in a 349-mm long by 85-mm thick annular shield of NaI(Tl) scintillation crystals which, together with the CsI(Na) portion of the phoswich and the CPD, forms the active anticoincidence shield for background rejection. The NaI(Tl) annular shield is divided into four optically-isolated quadrants or segments, each viewed by three 51-mm diameter photo-multiplier tubes.

The tungsten collimators are installed in the detectors so that OSSE detector motion scans the source with the collimator's 3.8 FWHM field-of-view. A background offset of 4.5 provides optimum modulation of the source contribution. This background offset, however, is programmable so that alternate background offsets can be selected in source-confused regions or for other scientific reasons. the nominal observing sequence observes the background on both sides of the source target. the average of these two backgrounds will generally be used in the data analysis as the "background" for the source observation. This technique minimizes the systematic effects of modulation of gamma-ray background local to the spacecraft. A typical detector observing sequence might consist of continuous repetition of the following sequence:

The motion occurs at a fixed rate of 2 per second. These observing programs can define sequences of up to 3 source and background offsets which can be used for source positioning or more complicated background modeling.

In the 90 between the spacecraft Z and X axes, all four detectors can be used for observations. Outside this range, the upper pair and lower pair of detectors begin to intrude into each other's fields of view. In that situation, the upper pair of detectors can be programmed to view one source and the lower pair can view another, independent target.

The ability to select viewing directions which contain interesting targets well away from the spacecraft Z-axis allows OSSE to make good use of the Z-axis earth occultation periods; by selection of the spacecraft orientation at the start of a two week observation, OSSE can move its detectors between the primary Z-axis target and the secondary target each orbit without reorientation of the spacecraft.

The experiment's data acquisition and control system incorporates varied modes of operation depending on the type of information desired during a particular observation. Diagnostic capabilities and redundancy are important features of the system; the ability to re-configure the experiment in the event of a failure has also been included. This versatility is achieved through the use of redundant programmable microprocessors.

Energy losses (spectra) in the phoswich are processed by three separate pulse-height and pulse-shape analysis systems for each detector. A low range pulse-height analyzer (PHA) covers the energy range 0.05-1.5 MeV; a medium range PHA covers the energy range 1-10 MeV; and a high range PHA covers energies greater than 10 MeV, nominally 10-250 MeV. Pulse-shape discrimination in the highest range is also used to separate neutron and gamma-ray energy losses in the NaI portion of the phoswich by utilizing the differing time characteristics of the secondaries produced by these interactions.

osse

Figure II-1

osse detector assembly

Figure II-2

Energy losses above 0.10 MeV in each of the four annular shield segments are detected and utilized for rejection of coincident energy losses in the phoswich. The energy losses in the 0.10-8 MeV range are also pulse height analyzed as part of a diagnostic calibration analysis system. The good spectral resolution of the shield segments (10.5% at 0.66 MeV) permit the use of shield spectra in the study of solar flares and possibly gamma-ray bursts. In addition, the low level discriminator event rates from the shields (set at 0.10 MeV, but programmable from 0.03 to 0.47 MeV) are processed by the OSSE central electronics for the detection and capture of gamma ray bursts. This gamma ray burst trigger operates independently from the trigger provided by the BATSE instrument.

The primary data from the OSSE instrument consist of time-averaged energy loss spectra from the individual detectors (spectral memory data). Each detector accumulates separate energy loss spectra from the validated gamma-ray events in each of the three phoswich energy ranges. The time interval for these accumulations, called the spectrum acquisition cycle time (TSAC), is a function of the overall operating mode of the OSSE experiment and is typically 16.384 or 32.768 seconds. The acquisition cycle time is controlled by the OSSE central electronics which selects the TSAC interval based on the OSSE telemetry format. Acquisition times in the range between 2.048 seconds and 32.768 seconds are available. The detectors' accumulation memories are double-buffered so that accumulation occurs in one memory while the other memory is being transmitted. The accumulation memories can support a maximum of 4096 events per spectral channel per TSAC.

Gamma-ray events for the two lowest phoswich energy ranges (0.05-1.5 MeV and 1-10 MeV) are separately accumulated into 256 channel spectra. These energy loss spectra have uniform channel widths of approximately 6 keV and 40 keV, respectively. the highest phoswich energy range (greater than 10 MeV) is processed to discriminate gamma-ray and neutron events. These gamma-ray and neutron events are then separately accumulated into 16 channel spectra. These energy loss spectra have uniform channel widths of approximately 16 MeV.

If high time resolution is required, e.g. for the analysis of fast pulsars, an alternate data mode is provided. In this mode (pulsar mode), spectral resolution or bandwidth is sacrificed in order to obtain time resolutions of up to 0.125 milliseconds. The pulsar mode processing permits the definition of up to eight energy bands to be included in the transmitted pulsar data. Gamma-ray events qualified as being in one of these eight energy bands are then processed in one of two pulsar modes:

1. event-by-event mode, where selected events are time-tagged and both energy loss and arrival time of the event are transmitted in the telemetry, or

2. rate mode, where high time resolution rate samples are taken in each of the eight energy bands.

The event-by-event mode of pulsar data provides the highest time resolution for the study of fast pulsars. This mode time-tags events accurate to 0.125 milliseconds at its highest resolution. Depending on the OSSE telemetry format, a maximum of approximately 290 events/second is supported in the event-by-event pulsar mode.

The pulsar rate mode can accommodate a much higher event rate but at the sacrifice of spectral resolution. This mode records the number of events in each of the defined energy bands at a specified sample frequency. The highest sample rate in this mode provides a resolution of 4 milliseconds. Sample times from 4 msec to 512 msec can be selected.

OSSE includes a gamma ray burst capture capability (burst mode) which is based on the measurement of the gamma-ray event rates in the NaI annular shields. These event rates are sampled by the OSSE central electronics on a selectable time scale of from 4 msec to 32 msec. A gamma ray burst trigger signal, either from BATSE or from internal processing, activates the storage of the next 4096 samples in a burst memory. A selectable fraction of this memory preserves the rate samples prior to the trigger. With coarser time resolution (approximately 1 second), rate samples from the individual NaI shield quadrants are available which can be used to identify the direction of the gamma ray burst. Selected shield spectra can also be accumulated using an on-board calibration and diagnostic PHA (CALPHA) in each detector subsystem.

The OSSE internal burst trigger monitors the annular shield rate samples for successive samples above a specified event rate as the indication of a gamma ray burst. The BATSE Burst Trigger Signal will be the primary burst trigger signal for OSSE since it incorporates more sophisticated trigger detection and a burst direction measurement which will be able to identify possible solar flares. The ability to react to solar flares will significantly enhance the OSSE science.

As an alternative to solar flare detection, the BATSE Burst Trigger Signal can be set to encode the results of an on-board calculation of the approximate position of a cosmic gamma ray burst. If the burst strength and position meet a set of criteria on rate, scan and azimuth angles (typically these are: rate greater than 40 counts per 64 msec, azimuth from the OSSE scan plane less than 11 degrees, and OSSE scan angle within the range of the instrument), the signal is sent to OSSE, which then enters a special burst mapping mode, typically mapping the region within 10 degrees of the reported position. This mode can be used to search for delayed emission from gamma ray bursts for a pre-determined period, typically 12 hours. If, in the meantime, an improved position becomes available via the BACODINE network, the OSSE mapping strategy can be updated appropriately, usually within one hour.

The OSSE instrument also carries a separate charged particle monitor (CPM) detector (see Figure II-1). This 19-mm diameter plastic scintillation detector is enclosed in a passive shield of approximately 3 g/cm2 of aluminum and is viewed by a 19-mm PMT. The event rates in this detector provide a monitor of the high energy charged particle environment for OSSE. Additionally, it provides a detection of the spacecraft entry into the South Atlantic Anomaly (SAA). The spacecraft is designed to turn off the OSSE detectors during traversals of the SAA; this charged particle monitor, however, remains on to provide integral charged particle dose monitoring for background modeling.

C. Instrument Parameters and Capabilities

Table II-1 provides a summary of the instrument parameters and capabilities of the experiment.

TABLE II-1

OSSE PARAMETERS AND CAPABILITIES

Detectors

1. Type: 4 identical NaI(Tl)-CsI(Na) phoswich, actively-shielded, passively collimated

2. Aperture Area (total): 2620 cm2 Photopeak Effective Area: see Figure II-3.

3. Field-of-View: 3.8 x 11.4 FWHM (0.05 - 10 MeV).

4. Energy Range: 0.05-10 MeV gamma rays (primary objectives), 10-250 MeV gamma rays (secondary objectives), 10 MeV solar neutrons (secondary objectives)

5. Energy Resolution: see Figure II-4.

6. Time Resolution: 16.384 or 32.768 sec in normal mode 0.125 msec in pulsar mode, 4 msec in burst mode

Source Exposure per Observation Period

An exposure time of 5 x 105 seconds can be achieved for a priority OSSE target viewed during a 14-day observation period with full telemetry coverage (this includes source and background observation)

Experiment 3- Sensitivity

1. Line Sensitivity (5 x 105 sec exposure; point source): see Figure II-5

2. Continuum Sensitivity (5 x 105 sec exposure): see Figure II-6 and Section VI of this Appendix

3. Gamma-Ray Burst Sensitivity: 1 x 10-7 erg cm-2

4. Solar Flare Line Sensitivity (103 s flare): (5-10) x 10-4 ph cm-2 s-1

5. Solar Flare Neutrons Sensitivity: 5 x 10-3 n cm-2 s-1

Pointing System

1. Type: Independent Single Axis

2. Drive System: Redundant Stepper Motor

3. Maximum Drive Speed: 2/sec

4. Range: 192 about the spacecraft Y-axis

(see figure II-1)

Compton Spacecraft-Experiment Interface Data

1. Weight: 1810 kg

2. Power Average: 192 Watts

3. Telemetry: 6492 Bits/sec

D. Data Reduction and Analysis

1. OSSE Data Types

The procedures required for analysis of OSSE data are ultimately determined by the instrument operating characteristics and the types of data available. As described above, data from OSSE consist of the following categories:

a. Time-integrated Spectra ("Spectral Memory Spectra")

OSSE internally accumulates spectra in the approximate energy ranges 0.05-1.5 MeV (low range), 1-10 MeV (medium range) and 10-250 MeV (high range), integrated over periods as short as 2.048 seconds or in multiples of 2.048 seconds. The low and medium energy-range spectra are sampled into 256 channels per detector, while the high-range spectrum is routinely sampled into 16 channels per detector (256 channels are available on longer time scales). In addition, the high energy range contains a 16-channel spectrum of >10 MeV neutrons. These spectra represent a large dynamic range and the best energy resolution attainable at the cost of moderate temporal resolution.

During ground data processing, spectral memory spectra are further summed into two-minute intervals corresponding to the OSSE source/background pointing periods. A two-week observation of a given celestial source could consist of as many as 10,000 of these basic two-minute intervals, or, after background subtraction, as many as 5000 separate source observations. For a typical source observation, the two-minute pointing interval will be the shortest time scale for which spectral memory spectra will be examined.

b. High Temporal Resolution Data ("Pulsar and Burst Data")

OSSE may also transmit data in one of several modes with temporal resolution finer than 2.048 seconds. The trade off is either reduced dynamic range (i.e., energy range), reduced energy resolution, or both. High time resolution data may be mixed with spectral memory data in the telemetry stream, resulting in further compromises in range, energy resolution and temporal resolution.

OSSE photopeak effective area for a single OSSE
detector

Figure II-3. OSSE photopeak effective area for a single OSSE detector (approximate). The effective area is plotted for (I) a 0 source offset exposure (source is at the center of OSSE's field of view), (ii) a 4.5 source offset exposure in scan angle (typically used for measuring the background), and (iii) the difference between the "source on" and "source off" exposure plots, which is the modulated effective area after background subtraction.

OSSE energy resolution

Figure II-4. OSSE energy resolution (approximate)

OSSE 3- line sensitivity for a point source

Figure II-5. OSSE 3- line sensitivity for a point source. The exposure time of 500,000 seconds includes half time on source and half time on background. This exposure time is typical of what can be achieved for a priority OSSE target viewed during a 21-day observing period with the current telemetry coverage efficiency.

OSSE 3- continuum sensitivity for a point source

Figure II-6. OSSE 3- continuum sensitivity for a point source. The exposure 500,000 seconds includes half time on source and half time on background. This exposure time is typical of what can be achieved for a priority OSSE target viewed during a 21-day observing period with the current telemetry coverage efficiency.

Pulsar data may consist of either "event-by-event" or "rate" modes. For "event-by-event" data, a time and energy are transmitted for each gamma-ray event and temporal resolutions of 0.125 milliseconds and 1 millisecond are available, with the lower temporal resolution yielding a larger energy range. For either temporal resolution, the energy range must be adjusted to limit the count rate to approximately 80 counts/second/detector maximum. Pulsar "rate" mode data consists of counts integrated into consecutive time intervals as short as 4 milliseconds but with limited energy resolution (maximum of eight energy bands).

Burst data, used for gamma-ray burst measurements, consist of 4096 samples of NaI(Tl) anticoincidence shield rates, with each sample representing a time interval as short as 4 milliseconds (yielding a total interval of 16 seconds) or as long as 32 milliseconds (yielding a total interval of 132 seconds). Shield burst data contain no energy information, with the exception of using the roving PHA (RPHA) to acquire limited spectra.

2. OSSE Data Product Characteristics

(See also Table V-2 in the Project Data Management Plan) The accessible data products are categorized as being Low Level or High Level depending on the degree of processing and the volume of the resultant product. The breakdown below includes the volume of data involved for Low Level products; the volume of High Level data is small relative to Low Level data.

a. Low Level Data Products

2-minute detector count spectra:

Approximately 10,000 intervals per 2-3 week observation period. Includes extensive header with instrument configuration, rates, and orbital environment. Primary spectrum data, calibration data, no diagnostic data. Volume: approximately 150 Mbyte per 2-3 weeks of observations.

Background-subtracted 2-minute count spectra:

Like the 2-minute spectra, noted above, but with a standard background subtraction algorithm applied, including propagated uncertainties. Volume: approximately 150 Mbyte per 2-3 weeks of observations (backgrounds and calibrations removed but includes uncertainties).

Pulsar data:

Event-by-event with resolution down to 0.125 milliseconds; limited band coverage. Rate with resolution down to 4 milliseconds; limited energy resolution. Volume: approximately 900 Mbyte per 2-3 weeks of observations.

Burst data:

4096 time bins from shields, Time resolution adjustable. Volume: approximately 20 Kbyte per burst.

b. High Level Data Products:

In general, all High-Level products will be presented for each of the four OSSE detectors.

Integrated Count Spectra:

These are background subtracted spectra summed over some or all of a given two week observation period (or multiple observation periods). They include header information listing instrument parameters, exposure, and integrated rates. Propagated uncertainties are included, but background spectra are not. Background subtraction is via the standard algorithm with no special corrections. Typically, data selection criteria will have been applied to the spectra to be summed. These criteria may include orbital environment parameters such as magnetic field values, time from SAA passage, etc., and/or pointing parameters such as earth horizon angle. Multiple spectrum sets with different selection criteria applied shall be available. These spectra will not have been corrected for instrument response. However, pulse height (PHA) calibrations (i.e., representative energies for each pulse height channel) will be supplied.

Integrated Incident Photon Spectra:

These spectra are similar to the integrated count spectra except that they have been corrected for instrument response. It should be emphasized that these are estimated incident photon spectra. The quality of the estimate (systematic uncertainty, resolution, compliance to models) depends on which of the several available instrument deconvolution algorithms has been used on the spectrum. Uncertainties included will reflect deconvolution errors, but will not necessarily indicate that some errors are correlated over large parts of the spectrum. Accompanying documentation will indicate algorithms used and point out possible problems.

Integrated Pulsar Light Curves:

Epoch-folded pulsar light curves integrated over selected portions of an observation shall be available in user-specified energy bands. These may be viewed as two-parameter spectra, as a function of energy and pulsar phase. Descriptive headers, exposure information per phase, and uncertainties will be included. Light curves as a function of trial period may also be supplied.

Energy Band Time Histories:

Time histories of the rates (corrected for exposure) in any energy band will be available. The time resolutions available depend on the instrument mode, and, at high time resolution, the energy band coverage may be limited. Multiple-correlated time series will be available as well (e.g., multiple energy bands, or energy bands and orbit environment data, correlated in time).

3. OSSE Data Analysis

The OSSE data analysis systems may be divided up into the following three areas for discussion:

a. OSSE Databases

The OSSE data analysis systems must be able to deal with diverse types of data, each of which is described by a large number of parameters and each of which can come from a wide variety of instrument operating modes. In addition, the volume of data of all types will be large. Spectral memory data alone will amount to approximately 100 Mbytes for a typical two-week source observation.

OSSE data will be stored in three database formats. All data will be available in an archived telemetry format. This format has no redundancy but also no compression and requires substantial analysis to generate usable forms.

Production analysis procedures will create three more refined databases. The spectral database will be the primary repository for spectra. Each spectrum, regardless of time scale, will include an extensive header containing most quantities required for further analysis (e.g., housekeeping, live times, etc.) integrated over the spectrum collection interval. This database may be viewed as the central interface for all OSSE processing. In order to facilitate access to individual spectra and groups of spectra, a separate catalog of spectral database headers will be maintained and managed by a formal database management system with an efficient query language and pointers to the spectral data itself.

In general, each scientist will have his/her own spectral database of spectra from the observation periods with which he/she is working. This database will not be sharable and, hence, the individual scientist will be free to modify it at will.

Production analysis also produces Pulsar Data Files (PDF's) containing high time resolution data together with some housekeeping information. These files are generally large and may reside in secondary storage.

In addition, calibration and instrument status data and results will be stored in an instrument model database. A certain degree of redundancy exists between this database and the spectral database header information, but the instrument model database maintains finer time scales and more information. In particular, each spectral database header will include a pointer to the instrument model database such that spectral data and calibration results may be associated. The instrument model database will include time variable information, such as energy calibrations, and fixed information such as simulation and preflight calibration results. The latter will be sharable among all users, while each user will generate his own copy of the former relevant to the observation period of interest.

b. OSSE Production Analysis

Production analysis consists of routine procedures performed on all data as it is received. These include reception, quality analysis, archiving, decommutation into separate data streams for various types of data, automatic summation into pointing intervals, database loading, automatic calibration and searching algorithm execution and hard copy record production. Production analysis will produce spectral database and instrument model database files on magnetic disk, optical disk and/or magnetic tape. These may then be accessed by the individual scientific user. A tape (or optical disk) library system will allow the user to select the volumes for the interval, observed source, etc., of interest. Production analysis functions will routinely be performed at the Naval Research Laboratory (NRL) within 24 hours of data reception. Archive and derived media will be shipped to Northwestern University (NU) on a regular, frequent basis.

Production analysis procedures will be largely transparent to the scientific user. A limited number of applications may require either special production runs, in particular if spectra on time scales faster than the pointing interval are desired, or specialized telemetry archive analysis routines. These routines, beyond a basic set, will be supplied on an as-needed basis.

c. OSSE Scientific Analysis

Scientific analysis consists of processes required for the scientific user to interact with the data with the goal of producing a scientifically usable result (e.g., estimates of production spectra from a source, models of source emission, etc.). Scientific analysis starts with semi-routine operations to remove instrumental effects. Foremost of these is background subtraction. Since all subsequent results are sensitive to errors in this procedure, it is included in the scientific rather than production analysis in order to give the scientific user more direct control over the process and to allow the user a feel for the compromises involved. As experience with OSSE data grows, background subtraction will become progressively more routine. However, certain situations (e.g., data near the limb of the earth), may always require special care.

Other scientific analysis procedures include unfolding the instrument response and examination of results for systematic effects and artifacts. The latter will be aided by the ability to select data sets from the spectral database while applying a wide variety of instrument state and environmental criteria. These procedures will include specialized algorithms for dealing with such situations as source confusion and extended source analysis.

As experience with the OSSE instrument has grown, revisions have been made to the energy calibration routines and response matrix generator. In some cases, usually bright and/or hard sources, this can have a significant impact on the analysis results.

Final results are achieved by a variety of processes such as data selection, combination and summation of spectra in a statistically meaningful manner, combining results from different pointing positions into sky maps, and modeling spectra and spatial distributions. Presentation and publication quality graphics utilities will be available to display results. The diverse processes that comprise scientific analysis will be tied together by a common interface known as the "Integrated GRO/OSSE Reduction Environment" (IGORE). This system will supply a common parameter and data interface to be used by all application programs. Based on the IDL data analysis language, IGORE has all of IDL's data manipulation and programming capabilities. Database, statistics, graphics and instrument specific applications will all be able to communicate with the user and with each other through IGORE. If the user wishes to write a specialized application, he only needs to know the IGORE interface rather than individual interfaces for each application and utility with which he must communicate.

Of course, this additional layer of software exacts a performance penalty, so in some compute-intensive applications, special purpose software may be written that by-passes IGORE and communicates directly with databases, graphics utilities, etc.

4. OSSE Data Analysis Facilities

Production analysis will take place at NRL only, but full scientific analysis facilities will be available at NRL, and the CGRO-SSC locations. IGORE accounts can be obtained by contacting the CGRO-SSC.

5. OSSE User Documentation

Detailed design documents for both hardware and software are currently available if required. Additional general documentation for new users of OSSE is available or in preparation; including an introduction to OSSE operating modes, data analysis procedures, database formats, production analysis procedures, and data analysis software.

a. GRO/OSSE Guest Investigator Guide - A basic introduction to OSSE -functions, data and terminology. Designed for use by persons totally unfamiliar with OSSE and with only minimal knowledge of gamma-ray observations. Numerous examples on using OSSE high-level data products. Available in PostScript format from CGRO World Wide Web homepage under OSSE and High-Level Products.

b. Mission Operations Guide - A description of OSSE operations and options, including OSSE operating modes, pointing capabilities, mission operations utilities and overall observing plan.

c. IGORE User Reference Guide - An encyclopedic reference to IGORE functions and procedures.

d. OSSE Instrument Paper - (Johnson, W.N, et al., 1993, ApJ Supp. 86, 693). A description of the calibration and operational characteristics of the OSSE instrument.

e. OSSE World Wide Web Server and Home Page - Brief OSSE description and OSSE papers available for downloading <http://osse-www.nrl.navy.mil/osse/osse.html>.

E. OSSE GUEST INVESTIGATOR OPPORTUNITIES

Because of the different operational and data analysis requirements of the Compton instruments, the Compton Guest Investigator program is tailored to the individual instruments rather than Compton as a whole. The complexity of the OSSE data analysis and the extreme care which must be exercised in interpreting results in a very low signal/background environment may require that the Guest Investigator initially work very closely with members of the OSSE Instrument Team or the OSSE Instrument Specialist. New Guest Investigators may find it useful to spend time at either NRL, NU or the CGRO-SSC for investigations involving extensive data analysis.

In accord with the original OSSE proposal, the OSSE team is retaining proprietary data rights and responsibilities to analyze and publish data on several selected topics or sources; these are:

a. New Galactic or local group supernovae (proprietary for data obtained up to two years after the event)
b. Comprehensive Galactic plane survey (not including proposed discrete source observations)
c. Comprehensive sky survey during Phases 3 and 4 (not including proposed discrete source observations)

Note that retention of exclusive data rights does not mean that Guest Investigator participation is excluded -- collaboration with the team is encouraged.

OSSE also offers a significant capability for observations of the sun. Solar array articulation and spacecraft thermal constraints require the Sun to be in the +X hemisphere of the spacecraft, OSSE's orientation capability often permits direct viewing of the Sun even when the other instruments are observing in some other direction. Thus, the Sun is often a candidate secondary target for OSSE viewing. Additionally, OSSE has a rather unique capability to quickly respond to solar flares based on the BATSE solar flare trigger signal and the offset pointing capability of the OSSE instrument. BATSE nominally takes 8 seconds to identify the transient event as coming from the solar direction. The OSSE response time to a solar flare would be about 60 seconds to re-orient the detectors to solar viewing. OSSE can also react to a BATSE solar trigger by successively accumulating spectra from the OSSE annular shield segments with a time resolution of up to approximately 4 seconds. This capability can provide modest spectral and temporal resolution data on solar flares without re-orientation of the OSSE detectors.

3. OSSE Data Archiving

OSSE will provide the Compton Science Support Center with reduced data sets within one year of processing the data to a usable format. This archival data, largely in FITS format, will be accessible by the community at large following NASA policy. Independent Investigators wishing to use archival data from OSSE will be assisted by the Compton Science Support Center.

F. OSSE Observation Definition Form

1. Defining an OSSE Observation.

Guest Investigators who are requesting data from an OSSE observation should fill out an Observation Definition Form. Hard copy of this form is included at the end of Appendix D of this NRA, however, the form must be included as part of the mandatory electronic proposal submission. The purpose of this form is for us to obtain from the proposer a definition of the data set needed for the planned investigation. If the form is inadequate for this purpose, the proposer should not feel constrained by its structure - text can be appended to the form under the special requirements box at the end. The instructions for filling out this form are as follows:

1. Use as many pages as you need to provide your target information. If the target is an extended source (> 1), the radius should be indicated.

2. In the "constraints" column, indicate any expected variability of the target, any spectral features to be studied, and the time resolution required for the observation (if less than 32.768 seconds). If the target is a pulsar, indicate the period. For spectral line observations, indicate the energy resolution required; refer to Figure II-4 for OSSE's limiting energy resolution.

3. Using the sensitivity plots provided in Figures II-5 and II-6, and the Guest Investigator's knowledge or model of the target, indicate the desired duration of the observation in weeks. The sensitivity plots are based on the typical exposure of 5x105 seconds possible during a normal 2 week observation of a priority OSSE target (includes source and background observation).

Note that with the expected 500-km orbit altitude during the 1997-2001 timeframe, a larger background count rate is expected. Predictions suggest a factor of ~1.5-2X current (November 1996) levels. To obtain the same detection sensitivity, OSSE observing time requests should be increased proportionately.

4. Indicate if there are special collimator field-of-view constraints that should be considered in planning the observation (for example, if the target is less than 5o from a potentially contaminating source).

5. Indicate any other special requirements (for example, increasing the normal gain of the phoswich; in this case indicate 2X gain in the mode column). Depending on the nature of these requirements, the Guest Investigator may wish to contact the OSSE team prior to proposal submission to discuss the feasibility of any special needs.

G. OSSE Data Product Availability

A summary of OSSE data formats, volume, and availability may be found in the GRO Project Data Management Plan, which is included as Appendix H of this NRA. See Tables V-1 and V-2 (p. 39, 45).

III. COMPTEL GUEST INVESTIGATOR PROGRAM

A. Introduction and Scientific Objectives

The Imaging Compton Telescope (COMPTEL) occupies the middle position (in both physical location and energy coverage) on the Compton spacecraft and surveys the gamma-ray sky in the energy range from 0.8 to 30 MeV. Measurements by COMPTEL effectively bridge the gap in observations previously existent between the hard X-ray and high-energy gamma-ray bands. COMPTEL combines a wide field of view (about 1 steradian) with good angular resolution. Its imaging properties have made it an ideal instrument for the first comprehensive survey of the sky at MeV-energies. With an energy resolution of 5% to 10% FWHM COMPTEL is also capable of conducting gamma-ray line spectroscopy. Measurements at MeV gamma-ray energies provide information critical to the detailed understanding of a wide range of high-energy astrophysical phenomena. The scientific objectives of the COMPTEL experiment can be grouped under the following headings:

Principal objectives:

Secondary objectives:

During Phase 1 of the Compton mission, COMPTEL conducted the first complete survey of the sky at MeV energies. From Phase 2 onwards selected celestial objects and regions are being studied in greater detail.

The Principal Investigator of COMPTEL is Dr. Volker Schönfelder of the Max-Planck-Institute for Extraterrestrial Physics (MPE), Germany. Co-Principal Investigators are located at the Laboratory for Space Research (SRON-Utrecht) (Dr. Wim Hermsen), Utrecht, The Netherlands, the Space Science Center of the University of New Hampshire (UNH)

(Dr. James Ryan), USA, and the Space Science Department, Astrophysics Division, of ESA/ESTEC (Dr. Kevin Bennett), Noordwijk, The Netherlands.

B. The COMPTEL Instrument

A detailed description of the COMPTEL experiment and its operating characteristics can be found in the report "Instrument Description and Performance of the Imaging Gamma-Ray Telescope COMPTEL aboard NASA's Compton Gamma Ray Observatory" by V. Schönfelder et al., 1993, ApJ Supp., 86, 657. Further general information on the COMPTEL instrument, as well as an electronic publications archive, can be found on the World Wide Web pages of the COMPTEL collaboration at <http://wwwgro.unh.edu/comptel/>. We provide in the sections below a summary description of the COMPTEL instrument and its properties.

1. Description

COMPTEL consists of two detector arrays. In the upper one, a liquid scintillator, NE 213A, is used and, in the lower, NaI crystals. Gamma rays are detected by two successive interactions: an incident cosmic gamma-ray is first Compton scattered in the upper detector, then totally absorbed in the lower. The locations of the interactions and energy losses in both detectors are measured. The accuracy in the measurement of these parameters determines the overall energy and angular resolution of the telescope. In some ways, COMPTEL is similar to an optical camera: the upper detector, analogous to a camera's lens, directs the light onto the second detector, comparable to the film, in which the scattered photon is absorbed. Although the photons are not focused, as in the case of an optical camera, data obtained by COMPTEL can be used to reconstruct sky images over a wide field of view with a resolution of a few degrees.

A schematic drawing of the telescope is shown in Figure III-1. The two detectors are separated by a distance of 1.5 m. Each detector is entirely surrounded by a thin anticoincidence shield of plastic scintillator which rejects charged particles. On the sides of the telescope structure, between both detectors, are two small plastic scintillator detectors containing weak 60Co sources, used for in-flight calibration of the instrument. Each calibration source consists of a cylindrical piece of 60Co-doped plastic scintillator of 3 mm thickness and 1.2 cm diameter viewed by two 1.9 cm-diameter photomultiplier tubes (PMTs).

The Upper Detector (D1) consists of 7 cylindrical modules of liquid scintillator NE 213A. Each module is 27.6 cm in diameter, 8.5 cm thick, and viewed by eight photomultiplier tubes. The total area of the upper detector is approximately 4188 cm2. The Lower Detector (D2) consists of 14 cylindrical NaI (Tl) blocks of 7.5 cm thickness and 28 cm diameter, which are mounted on a supporting base plate. Each block of NaI is viewed from below by seven photomultiplier tubes. The total geometrical area of the lower detector is 8620 cm2. Each anticoincidence shield consists of two 1.5 cm-thick domes of plastic scintillator NE 110, with each dome viewed by 24 photomultiplier tubes. With the exception of the front-end electronics, all detector electronics are mounted on a platform outside the detector assembly.

2. Instrument Modes for Data Acquisition

The Instrument Ground Support Equipment (IGSE) for COMPTEL is located at the University of New Hampshire, USA. All commands sent to the instrument originate at the University of New Hampshire, which also receives daily "quick-look" data during real-time passes of the spacecraft. These data are inspected immediately upon receipt in order to verify the satisfactory health, safety, and performance of the instrument.

a. Primary Modes for the Acquisition of Scientific Data

1) Double-Scatter Telescope Mode

COMPTEL operates primarily as a double-scatter gamma-ray telescope. In this mode an incident cosmic gamma ray is electronically identified by a delayed coincidence satisfying time-of-flight criteria between the upper and the lower detectors, combined with the absence of a signal from all charged particle shields and from the calibration detectors. Gamma-ray events are initially screened on board the spacecraft according to coarse pulse-height, time-of-flight, and charged-particle veto limits, which are applied to reject background events. The data are transmitted in telemetry packets where events satisfying all selection criteria (referred to as "Gamma 1" events) have first priority for transmission in the telemetry stream, followed by events selected according to a secondary set of onboard criteria ("Gamma 2" events). The telemetry rate of 6125 bits per second permits the transmission of approximately 20 events per second from the spacecraft.

The quantities measured for each selected gamma-ray event are:

(1) The energy of the recoil electron of the Compton scattered gamma ray in the upper detector Ee' (Ee' > 50 keV), determined from the summed pulse heights of the eight PMTs associated with the triggered D1 module;
(2) The location of the interaction in the upper detector, determined from the relative pulse heights of the eight PMTs associated with the triggered D1 module;
(3) The pulse shape of the scintillation in the upper detector, provided by the output of a pulse-shape discriminator circuit;
(4) The energy loss Ee" in the lower detector ( Ee" > 500 keV), determined from the summed pulse heights of the seven PMTs associated with the triggered D2 module;
(5) The location of the interaction in the lower detector, determined from the relative pulse heights of the seven PMTs associated with the triggered D2 module;
(6) The time-of-flight of the scattered gamma ray from the upper to the lower detector; and
(7) The absolute time of the event.

The energy of an incident cosmic gamma ray is estimated by summing the energies deposited in the upper and lower detectors, assuming total absorption of the scattered photon. From the energy losses and the interaction locations recorded for both the upper and lower detectors the arrival direction of an incident gamma ray is calculated by application of the Compton scattering formula. For a true Compton scatter event the arrival direction of a cosmic gamma ray is known to lie on a so-called "event circle" on the sky. The center of the circle is the direction of the scattered gamma-ray within the telescope, and its angular radius is derived from the energy losses in both detectors. In the ideal case of totally-absorbed Compton-scattered photons, the intersection of many such event circles on the sky yields the location of the gamma-ray source. In actual practice a detailed understanding of the properties of the scintillation detectors, and of the response of the telescope to background, multiple interactions, and partial absorption events, must be exploited to determine accurate source locations.

2) Burst Mode

In addition to the normal double-scatter telescope mode of operation, two of the NaI crystals in the lower D2 detector assembly of COMPTEL are also operated simultaneously as burst detectors. These two modules are used to measure the time history and energy spectra of cosmic gamma-ray bursts and solar flares. The burst detector modules are equipped with a dedicated electronic subsystem, the Burst Spectrum Analyzer (BSA). One of the burst modules provides energy coverage over a low range (~100 keV to ~1 MeV), and the other over a high range (~1 MeV to ~10 MeV). The burst detectors are sensitive to incident radiation over 4 steradians, though the field of view not obstructed by other portions of the spacecraft totals approximately 2.5 sr.

The burst system operates in three different internal modes. The background mode is the usual operating mode: spectra from both burst detectors are accumulated over a tele-commandable period of time (from 2 s to 512 s) and read out continuously or at a reduced rate (also tele-commandable, e.g., every 5 min). These data are used to establish the celestial and instrumental background before and after a burst event. The BSA switches into burst mode upon receipt of a trigger signal from the BATSE experiment. A total of six spectra per energy range are accumulated with an integration time that can be set from 0.1 s to 25.6 s. After completion of the sixth spectrum, the BSA enters a tail mode where up to 256 spectra per energy range with tele-commandable integration times from 2 s to 512 s are accumulated and stored for transmission. The tail mode is of value for longer-lasting

schematics of COMPTEL

Figure III-1

burst events (e.g., solar flares). After completion of the tail mode, the BSA resumes the background mode and is ready to receive the next burst trigger.

3) Solar Modes

i. Solar Gamma-Rays

In the event of a solar flare, COMPTEL can observe solar gamma rays in both the double-scatter telescope mode, provided the Sun is within the field of view of the instrument, and in the burst mode. Telescope observations of solar gamma rays have the advantage of a known source position, thus eliminating the need in later analysis to consider events that do not originate from the direction of the Sun. Upon receipt of a burst trigger from BATSE solar gamma rays can also be recorded by the two burst detectors aboard COMPTEL. The two burst modules, due to their rapid accumulation rates, can provide information on the initial fast burst of gamma rays from a solar flare.

ii. Solar Neutrons

In addition to observing gamma rays from a solar flare, COMPTEL is also capable of detecting solar neutrons. Neutron interactions within the instrument occur when an incident solar neutron elastically scatters off a hydrogen nucleus in the liquid scintillator of an upper D1 module. The scattered neutron may then interact and deposit all or a portion of its energy in one of the lower D2 modules, providing the internal trigger signal necessary for a double scatter event. The energy of the scattered neutron is deduced from its time of flight from the upper to lower detector, which is summed with the energy measured for the recoil proton in the upper D1 module to obtain the energy of the incident solar neutron. The computed scatter angle of the neutron, as with gamma rays, yields an event circle on the sky, which can be further constrained since the true source of the detected neutrons is assumed to be the Sun.

In practice, neutron observations are conducted in the following manner: within two minutes of an initial burst trigger, BATSE sends a second signal to COMPTEL indicating that the burst originates from the general direction of the Sun. COMPTEL can then automatically be commanded to enter an alternate event selection mode to measure solar neutrons for a period of 90 minutes (or approximately one orbit of the spacecraft). This is achieved by shifting the acceptance window for the time of flight of particles from the upper to the lower detector to allow for the slower-moving neutrons, compared to the speed-of-light gamma rays. Gamma ray events continue to be accumulated simultaneously with the neutrons; the two types of particles are later distinguished by their respective time-of-flight and pulse-shape signatures.

b. Secondary Modes for Data Acquisition

In addition to those instrumental modes intended for the acquisition of scientific data, the following modes of operation can also be specified. These secondary modes are intended for the collection of data relevant to in-flight calibration, the assessment of instrument performance, and for the study of background effects due to instrumental activation and atmospheric albedo.

1) SAA Mode ­ During passages through the region of the South Atlantic Anomaly (SAA) the high voltages are switched off to prevent damage to the photo-multiplier tubes. The period of SAA transits can be used to verify the pedestals of the on-board pulse height analysis system and to test the burst analysis system.

2) Proton Calibration Mode ­ In this mode data can be obtained to investigate the instrument response to charged particles, and to evaluate the performance of the anticoincidence veto domes in the suppression of the background due to charged particle interactions.

3) Upper Detector (D1) Single Event Calibration Mode ­ Energy spectra of the interactions occurring in the D1 modules are obtained, in addition to normal gamma-ray telescope events. Spectral lines of background interactions can be used to verify the calibration of individual D1 modules. This calibration mode is typically invoked during earth occultation of the primary scientific viewing target.

4) Lower Detector (D2) Single Event Calibration Mode ­ Similar to (3) above, but for the D2 modules.

5) Atmospheric Neutron Mode ­Similar to the Solar Neutron Mode, described in the previous section, but on command only; intended to investigate instrumental background due to neutrons from the Earth's atmosphere.

C. Instrument Parameters and Capabilities

A summary of the principal operating characteristics of COMPTEL is given below in Table III-1 (see also section III.E.2 of this Appendix).

TABLE III-1

COMPTEL CAPABILITIES AND OPERATING CHARACTERISTICS

FOR GAMMA-RAY OBSERVATIONS

a. Detector type:

1) upper detector D1: liquid scintillator NE 213A

2) lower detector D2: NaI (Tl)

b. Geometric arrangement:

1) D1: 7 cylindrical modules 27.6 cm in diameter and 8.5 cm deep; total geometric area 4188 cm2

2) D2: 14 cylindrical modules 28 cm diameter and 7.5 cm deep; total geometric area 8620 cm2

c. Effective area for gamma-rays: 20 - 50 cm2 (no event selections applied to the data)

d. Energy range: 0.8 - 30 MeV

e. Energy resolution: 5 - 8% (FWHM)

f. Angular resolution: 1.7 - 4.4 (FWHM)

g. Geometric factor: 5 - 30 cm2 sr

h. Field-of-view: ~ 1 sr

I. Accuracy of Source Position Determination: 5 - 30 arcmin

2. Experiment Sensitivities:

a. Telescope Observations of a Point Source

1) Minimum source detectability at 3s, 1-30 MeV (2-week observation) = 1.6 x 10-4 cm-2 s-1

(See E.2. and Section VI for additional detail)

2) Line sensitivity (3, 2-week observation) = 6 x 10-5 cm-2 s-1 at 1 MeV = 1.5 x 10-5 cm-2 s-1 at 7 MeV

b. Burst Observations

1) Telescope mode lower limit: S(>1 MeV) = 2 x 10-6 erg cm-2 ,upper limit: S(>1 MeV) = 1 x 10-4 erg cm-2 (45 s duration)

2) Single detector burst mode variable, depending on duration (see section III.E.2.b, Tables III-3, III-4).

c. Solar Flare Observations

1) Telescope mode = 1.2 ph cm-2 (1-3 MeV, 100 s duration) ,= 0.5 ph cm-2 (3-10 MeV, 100 s duration)

2) Single detector burst mode as for burst observations, above

3. Miscellaneous Instrument Specifications:

a. Weight: 1460 kg

b. Dimensions: 2.61 m x 1.76 m diameter

c. Power: 206 W

d. Telemetry Rate: 6125 bit/s (equivalent time-average)

e. Timing accuracy: ±1/8 msec with respect to UTC.

D. Data Reduction and Analysis

COMPTEL is a multi-detector instrument consisting of 21 main detector elements (7 upper and 14 lower modules) and their 154 photo-multiplier tubes; each possible pairing of an upper and lower detector can be treated as a separate telescope system, yielding (7 x 14 =) 98 "mini-telescopes." Auxiliary to the main detectors are the four veto domes and two small scintillator detectors containing radioactive sources used for in-flight calibration of the instrument. The raw data consist of individual "event messages" containing approximately 20 parameters which characterize a selected event once the telescope is triggered by an incident gamma photon. The COMPTEL database consists primarily of large sets of such event messages, in addition to a variety of calibration and in-flight orbital and "housekeeping" datasets which are maintained to monitor instrument performance and to correct for gain shifts and drifts in the operation of various detector elements over the course of the mission.

To process and analyze the enormous quantities of data received from COMPTEL, the collaboration has developed a customized and modular data analysis package, COMPASS ("COMPTEL Processing and Analysis Software System"), which uses the ORACLE database management system for configuration control and to monitor user access to the data. Each of the four collaborating institutions is responsible for the development and maintenance of particular software subsystems which make up the COMPASS package. COMPASS is designed to run at all four collaborating institutions with approximately equal capabilities for data processing and scientific analysis, though particular processing tasks have been assigned within the collaboration to specific sites.

The routine processing and analysis of data received from COMPTEL proceeds in three major steps: (1) the processing of raw data, (2) instrument response, background, and exposure determination, and (3) scientific analysis. For a further introduction to COMPTEL data analysis procedures, refer to the reports by Diehl et al. (1992) and den Herder et al. (1992), in the proceedings of the Compton Observatory Science Workshop (NASA CP-3137).

1. Routine Processing of Raw Data

The routine, "production" processing of raw data from COMPTEL is carried out at one central site, the Max-Planck-Institute in Germany. Once data tapes are received from the Packet Processor (PACOR) facility at NASA Goddard Flight Center, the first priority is the decoding and sorting of the telemetry data into various data categories (event messages, burst system spectra, instrument status data, in-flight calibration data, and housekeeping data). The next major task is to establish the correction factors for the gain fluctuations of the 154 photo-multiplier tubes (PMTs) associated with the upper and lower detector modules, using in-flight calibration data. Energy loss values are then assigned to each detected interaction, and the locations of gamma-ray interactions within the individual detector modules are determined, using pre-existing reference tables derived from the analysis of pre-launch calibration datasets. From these values the direction of the scattered photon is computed in celestial coordinates, and the angle of scattering calculated. The resultant values for each calibrated event, along with its absolute time of occurrence and associated time-of-flight and pulse-shape information, is written to a database of processed events, and constitutes "low-level" processed data.

2. Determination of the Instrument Response, Background, and Effective Exposure

Scientific analysis requires an accurate knowledge of the instrument response (i.e., COMPTEL's efficiency in detecting incident radiation), of the background contained in the dataset of accumulated events, and of the effective exposure of the telescope during a given observation period. Detailed knowledge of the background level is obtained from data taken in special instrument modes, by comparison to pre-launch calibration data, and by comparison to data produced in Monte Carlo simulations based on a model of the COMPTEL instrument. The fraction of background events can be significantly reduced by applying certain selection criteria to individual events within the dataset (e.g., time-of-flight and pulse-shape windows). The optimum setting of these event acceptance limits, and the choice of "standard" selection criteria, requires an exhaustive and detailed analysis of all recorded events.

The exposure factor of the telescope for a given observation period is the product of the observation time and the telescope response. The telescope response is an extremely complex function, which depends both on the energy and arrival direction of the incident gamma-rays, and on the range of accepted scatter directions and derived scatter angles. The response of the telescope changes with time since, at any given moment in orbit, the subset of COMPTEL's (7 x 14 =) 98 mini-telescopes exposed to the high radiation of the Earth's horizon will vary. Similarly, if the energy thresholds of the detectors or the acceptance criteria for events are altered, then the overall response of the telescope will change. A software module has been developed within COMPASS to calculate the instrument response for each possible configuration of the telescope, based on the results of pre-launch calibration, Monte Carlo simulations and an analytic understanding of the telescope. Finally, the observation time must be corrected for instrument dead-time effects in order to obtain the true live time for a given observation period; again, it will be different for each mini-telescope. The final product of this analysis effort is a set of matrices whose contents characterize the response, background, exposure, and overall sensitivity of the instrument for a given observation period and field of view. This collection of datasets will be used in further scientific analysis, and constitutes the first stage of "high-level" processed data.

3. Routine Scientific Analysis

Routine scientific analysis of processed COMPTEL data begins with the production of images of the sky over the field of view of the instrument for a given observation period. These skymaps are generated using a "standard" set of selection criteria for accepted events, and the instrument response and exposure datasets appropriate for that particular observation period. Model distributions of source and background counts are convolved with the instrument response, and a map which best fits the observed data is produced. A number of deconvolution algorithms can be employed within the COMPASS environment to produce images of the sky; these are based primarily on maximum entropy techniques, or on one of several event projection methods. Each skymap is searched for gamma rays using, for example, maximum likelihood and cross-correlation techniques, and the parameters of an identified source are determined. Physically extended sources (e.g., due to molecular clouds) and regions of diffuse gamma-ray emission are also identified by means of analysis techniques developed within COMPASS for that purpose. If known or suspected pulsars, or other sources with expected temporal signatures, lie within the field of view of the instrument, a search for modulated emission can be performed. Similarly, the analysis of burst events, and of gamma rays and neutrons of solar origin, is carried out by specialized tasks within COMPASS. The scientific data which result from these analyses are considered "high-level" processed data.

Detailed examination of these "standard" scientific data products may indicate that further data processing for particular regions or sources is warranted. Selection criteria for accepted events may be altered to optimize the signal-to-noise ratio, or it may be deemed necessary to produce a new set of response, exposure, or background matrices more appropriate to the circumstances of a given observation. Several iterations of the analysis procedure described in the preceding sections may be required before consistent and satisfactory results are obtained. In the course of periodic team meetings the COMPTEL collaboration reviews all routine analysis procedures, and certifies the scientific integrity of "low-level" and "high-level" data products obtained to date.

4. Categories of COMPTEL Data Products

The general characteristics of COMPTEL data products are outlined below with recommendations regarding their use by Guest Investigators. A more detailed description of COMPTEL data products can be found in the document "The Gamma Ray Observatory Project Data Management Plan" (PDMP=Appendix H of this Research Announcement).

COMPTEL data products are grouped into three broad categories reflecting the level of data processing up to that stage of the analysis:

a. "Raw data" include all instrumental effects, uncorrected for drifts in gain, etc. These data are essentially reblocked telemetry data.

b. "Low-level data" are "calibrated" datasets in which instrumental effects have been removed, and a conversion from "engineering" to "physical" units has been accomplished (e.g., pulse-height channel number to energy). These data are primarily processed event lists, along with associated instrument status and housekeeping files.

c. "High-level data" result from the scientific analysis of low-level data; a wide range of data products is possible, as outlined in the next section. Two general sub-catagories of high-level data are possible: one involving data where "standard" event selection criteria have been employed (for initial scientific interpretation, and assessment of data quality for further scientific analysis), and another where modifications to the standard sets of selection criteria may be used to obtain a final, fully-optimized scientific data product.

Typical COMPTEL data products available to Guest Investigators will be:

Low Level: Event data files (calibrated), burst raw spectra (calibrated), orbit and aspect data tables, instrument mode tables, and instrument performance summaries. These data permit some timing analysis (e.g., for known pulsars), the study of instrumental background, and possibly some limited analysis of strong sources. Detailed study of individual sources is problematic at best using low-level data, since exact source locations cannot be unambiguously assigned at this stage of data reduction; events from sources as far apart as 30 on the sky may populate the same regions in COMPTEL dataspace (i.e., photons incident on the telescope from different parts of the sky may produce nearly identical measured event parameters). Only after proper deconvolution with the instrument response function is it possible to accurately separate the multiple sources which may be present within COMPTEL's wide field of view.

High Level: The high-level data products immediately available to Guest Investigators will normally be those where "standard" event selection criteria have been applied in the data processing and analysis. Such data products would include: maximum entropy skymaps, source parameter tables (e.g., source positions and fluxes with confidence intervals), the binned event, background and response matrices used in the skymap and source analyses, pulsar light curves, and deconvolved energy spectra. These data are appropriate for direct scientific interpretation. The event, background, and response matrices may be used to perform an independent analysis of COMPTEL data, and to undertake comparisons with data taken at other wavelengths. However, this is not a simple task due to the multi-dimensional event signature and the highly non-diagonal response matrix of COMPTEL. Standard analysis techniques and software packages appropriate for other spectral regimes may require extensive modification before they can be applied to COMPTEL data.

It is the recommendation of the COMPTEL team that:

Guest Investigators intending to use low-level data products should be prepared to invest an extended period of time learning the details of instrument operation and data processing procedures. This can most efficiently be done at one of the COMPTEL institutions.

Guest Investigators requesting high-level data products must decide if "standard" COMPTEL images, spectra, etc. are sufficient for their purposes, or whether the required degree of sensitivity demands an independent re-analysis using the event data, background, and response matrices that would be provided. The latter approach would require a minimum of several weeks of in-depth study of COMPTEL data processing procedures and related instrumental subtleties.

E. COMPTEL Guest Investigator Opportunities

1. Overview

Research proposals for the COMPTEL Guest Investigator program re solicited from outside scientists through this NASA Research Announcement, and a similar one released by the German Ministry for Research and Technology (BMFT). Selected investigators may wish to spend time at one of the collaborating COMPTEL Institutions or the Compton Observatory Science Support Center.

Some scientific objectives are not suited for a Guest Investigator program. These include, for example, the production of an all-sky gamma-ray map, and the establishment of a source catalogue. Such items require the analysis of the entire COMPTEL data set and are reserved for the instrument team. Refer to the section on "Proprietary Data Rights" in Appendix C of this Research Announcement.

2. Sensitivity Estimates

To assist potential Guest Investigators in assessing the feasibility of a proposed observation preliminary sensitivity estimates are provided below for each of the main scientific observing modes (see also Table III-1).

a. Telescope Observations

1) Point Sources of Continuum Emission

An expression for the minimum detectable flux, Fs min, from a point source at n standard deviations above some background level, over a specified energy range, is given by

Fs min = n (NB)½ / (A D) (T T) ph cm-2 s-1 MeV-1

where

n = "confidence" level

(NB)½ = uncertainty in the total background (ph MeV-1)

A = detector area (cm2) = area of D1 modules for COMPTEL

D = detector efficiency (a function of photon energy and incidence angles)

T = total time of observation (sec)

T = observing efficiency over the time T and

NB = total background counts

FB = background flux (ph cm-2 s-1 sr-1 MeV-1)

The product of detector area and detector efficiency is defined to be the "effective area," Aeff, and the product of observing time and observing efficiency the "effective observing time," Teff. The product (Aeff Teff; units: cm2 s) is termed the "effective exposure" of the instrument for a given observation.

In Table III-2 below we provide estimates of the source flux detectable by COMPTEL in double-scatter telescope mode over four energy ranges; these values correspond to a 3 detection of an on-axis source. An observing efficiency of T = 0.3 was assumed (allowing for earth occultation of the source, SAA passages, etc.), yielding Teff = 4 x 105 seconds for a typical two-week observing period. The number of background events was estimated by assuming a 3-wide (±1.5, = radius of the angular resolution over a specified energy range) acceptance interval around a celestial gamma-ray source positioned exactly on the telescope axis.

The following estimates of instrument sensitivity are based on pre-launch calibration data, Monte Carlo simulations, and actual measurements of the instrumental background observed in flight by COMPTEL. The background rate can be influenced in a very sensitive way by event selection criteria. The most critical selection parameters are the energy thresholds in D1 and D2, the time-of-flight window for accepted events, and the rejection criteria for earth albedo events. Based on experience acquired during the Phase 1 Sky Survey of the Compton mission, an initial set of selection criteria has been defined to optimize the scientific content of the data received from COMPTEL. These selections may be further modified in the future.

TABLE III-2

COMPTEL 3 POINT SOURCE CONTINUUM SENSITIVITY

(FOR A TWO WEEK OBSERVATION)

Energy Angular Acceptance Effective NB Fs min(3)

Resolution Radius for Exposure

Background Aeff Teff

Events

(Mev) (1s - deg)   (deg)   (cm2 s) (ph) (cm-2 s-1) 

0.75 - 1  2.0 3.0  24.4 x 105   9570  1.2 x 10-4 

1 - 3   1.67 2.5   35.6 x 105  50920  1.9 x 10-4 

3 - 10   1.2 1.8   49.2 x 105  16870  7.9 x 10-5 

10 - 30  1.0  1.5  63.6 x 105  1150  1.6 x 10-5 
1 - 30                               1.6 x 10-4 

Note that these values reflect the response of COMPTEL to an on-axis source. Observing efficiency declines with increased zenith angle. Figure III-2 illustrates the dependence of detection efficiency on the zenith angle of a source at a representative energy.

2) Point Sources of Line Emission

An analysis similar to that outlined above can be performed for sources of line emission. The 3-sigma sensitivity of COMPTEL to such mono-energetic emission ranges from approximately 6x10-5 cm-2s-1 at 1 MeV, to approximately 1.5x10-5 cm-2 s-1 at 7 MeV. COMPTEL is slightly less sensitive to cosmic line emission in the vicinity of two prominent instrumental background lines at 2.2 and 4.4 MeV.

b. Burst Observations

We provide below a summary of sensitivity estimates for the detection of cosmic gamma-ray bursts by COMPTEL. Refer to the COMPTEL calibration paper of Schoenfelder et al. (1994, ApJ Supp., 86, 657) for additional details and references.

1) Double-Scatter Telescope Mode

In general, the angular resolution of a point source observed with the double-scatter telescope mode of COMPTEL is determined by the spatial and energy resolution of the upper and lower detectors. If S(>E) is the burst fluence (erg cm 2), then the lower detection threshhold is given by

S(>1 MeV)low = E x N(>1 MeV)/Aeff

where E is the mean photon energy (5 MeV) over the energy range 1 to 30 MeV, assuming a power-law spectrum, and Aeff is the effective detection area for totally absorbed on-axis events. If we require N(>1 MeV) = 30 events and Aeff = 30 cm2, we obtain S(>1 MeV) = 8 x 10 6 erg cm 2. The number of background evenets within a 4 x 4 degree resolution element of COMPTEL is about 0.2 counts s 1 (see Table III-2). Therefore the detection of gamma-ray bursts is photon limited, not background limited.

An upper limit to the burst-detection threshhold is defined by the rate of coincidence-by-chance events between the COMPTEL upper and lower detectors. This chance-coincidence rate has been estimated to be approximately 2000 counts s-1, for a gamma-ray burst of typical duration (45 s) and spectral shape (S(>20 keV) = 3 x 10 3 erg cm 2). Therefore, for such typical burst parameters, COMPTEL is sensitive to those burst events with S 10 4 erg cm 2.

For the 4-year period April 1991 to April 1995, COMPTEL has detected 28 cosmic gamma-ray bursts within its field of view of one steradian, in reasonable agreement with pre-launch estimates. Depending on the burst fluence, burst locations can be determined to an accuracy of approximately one degree at MeV gamma-ray energies. Further summary results on the detection of gamma-ray bursts within the field of view of COMPTEL can be found in Kippen et al. (1995, Proc. 24th ICRC (Rome), 2, 61), and Hanlon et al. (1994, A&A, 285, 161).

2) Single Detector Burst Mode - Continuum Sensitivities

Using a background spectrum estimated by the BATSE team, 5- continuum sensitivities of burst fluence S (erg cm 2) for 4 energy ranges and 4 typical burst durations have been calculated for the single-detector burst modules of COMPTEL. Mean photon energies per energy range were derived using typical continuum spectral shapes for burst events: a thermal bremsstrahlung model (exp{-E/E0}; E0 = 250 keV) for energies below 300 keV, and a power law model (E 2) for energies above 300 keV. The number of burst photons were calculated following Li and Ma (1983 Ap. J. 272, 317): S = (Non - a x Noff)/[a x (Non + Noff)]1/2

where the significance S = 5 sigma; number of burst photons Ns = Non - a x Noff; Non = source counts plus background counts accumulated during ton; Noff = background counts accumulated during toff; a = ton/toff with toff = 100 s and ton = 0.1 s, 1 s, 10 s, and 100 s of burst duration. The 5- threshholds on burst fluence are shown in Table III-3.

3) Single Detector Burst Mode - Line Sensitivities

The sensitivity of the COMPTEL burst modules to possible line features in the spectra of cosmic gamma-ray bursts has been estimated, based on pre-CGRO reports, for a range of potential line strengths and burst durations. Refer to the COMPTEL calibration paper of Schoenfelder et al. (1994, ApJ Supp., 86, 657) for a description of these sensitivity estimates and their derivation.

TABLE III-3 CONTINUUM SENSITIVITIES (5 ) TO BURST FLUENCE S (x 10-6 erg cm-2) FOR VARIOUS ENERGY RANGES AND BURST DURATIONS (ton) Energy < E > S Range (x 10-6 erg cm-2)

(MeV) (MeV) ton = 0.1 s ton = 1.0 s ton = 10 s ton = 100 s 
0.1 - 0.3   0.19       0.113      0.360      1.19      5.08
0.3 - 1.0   0.55       0.158      0.502      1.66      7.11
1 - 3       1.70       0.264      0.839      2.78      11.9
3 - 10      5.40       0.748       2.38      7.90      34.1
A plot of COMPTEL detection efficiency, in double-scatter
telescope mode, as a function of the zenith angle of the incident gamma-ray
photon, at a representative energy of 1275 keV.

Figure III-2. A plot of COMPTEL detection efficiency, in double-scatter telescope mode, as a function of the zenith angle of the incident gamma-ray photon, at a representative energy of 1275 keV. The upper curve is without any software selections on the data; the lower curve with "standard" event selection criteria applied.

c. Solar Observations

1) Gamma-Ray Observations

Observations of gamma-rays from solar flares can be undertaken in both the double-scatter telescope mode and the burst mode of COMPTEL . Sensitivity limits in general will be comparable to those outlined above for gamma-ray bursts. Preliminary analysis of COMPTEL flight data indicates that, for a solar flare of 100 s duration, the COMPTEL sensitivity in telescope mode to solar gamma radiation is approximately 1.2 ph cm-2 (1 - 3 MeV) and 0.5 ph cm-2 (3 - 10 MeV). An added complication for solar observations is that elevated fluxes, from even moderately strong flares, can result in severe instrumental dead times. Absolute flux values can therefore be extremely difficult to determine.

2) Neutron Observations

Initial analysis of COMPTEL flight data for solar flare observations indicates a 5 sensitivity to solar flare neutrons, above the quiescent solar neutron background, of approximately 1.0 x 10-3 n cm-2 s-1 over a neutron energy range of 10 - 150 MeV. For a duration of 1800 s, this translates to an integrated neutron fluence of 1.8 n cm-2. At present, an unknown factor in the neutron sensitivity is the dead time experienced by the telescope in the later phases of a solar flare. Initial observations by COMPTEL of solar flare neutrons have been affected significantly by the pulse pile-up of thermal x-rays in the forward charged particle shields of the telescope. Even though the first possible detection of neutrons from a flare occurs more than 8 minutes after the onset of the gamma-ray "flash," thermal x-rays persist for considerably longer periods, negatively influencing the telescope's ability to register solar neutrons; elevated and extended fluxes of solar gamma rays can also contribute to the instrumental dead time. Modifications to the solar neutron observation mode of COMPTEL have been implemented to minimize the effects of thermal x-rays.

3. Parameters Required for the Specification of an Observing Program

A potential Guest Investigator who wishes to propose for observing time with COMPTEL should provide the following information on the observation definition form(s) included with this Research Announcement. The primary requirements are outlined below. Note that there are relatively few instrumental configurations that can be selected; detector energy thresholds and the settings for time-of-flight and pulse-shape windows, for example, will, in general, already be optimized for each possible observing mode.

a. Target Name and Instrument Pointing. Provide the name and type (pulsar, AGN, etc.) of the desired target of the observation, along with its position in right ascension and declination epoch 2000.0). For multiple targets within a given field of view, specify the desired center of the field.

b. Energy Range (MeV). Specify the energy range of interest for this observation.

c. Duration of the Observation. An estimate of the number of days of on-axis observation required.

d. Observing Mode. Specify one of the standard scientific observing modes of COMPTEL: double-scatter telescope mode, burst mode, solar gamma or neutron mode.

e. Special Operational Requirements. Indicate "non-standard" observational requirements, if any, for the observing mode specified (e.g., threshold settings, accumulation rates for bursts)*. f. Special Scheduling or other Time Constraints. Indicate if special scheduling constraints are required (e.g., time of year, simultaneous with observations by other instruments, etc.).

g. Standard Data Products. Indicate the standard data products desired for this observation.

h. Special Data Products. Indicate if any "non-standard" data products are required for this observation. Note that special data processing requirements may delay the delivery of data products to the Guest Investigator.*

*N.B. Subject to technical review/feasibility study by the instrument team.

F. COMPTEL Facilities for Guest Investigators

Any of the four collaborating COMPTEL institutions may serve as a facility for the pursuit of a guest investigation: the Max-Planck-Institute for Extraterrestrial Physics (MPE), Garching, Germany, the Laboratory for Space Research (SRON-Utrecht), Utrecht, The Netherlands, the Space Science Center of the University of New Hampshire (UNH), Durham, New Hampshire, USA, or the Space Science Department of ESA/ESTEC (SSD), Noordwijk, The Netherlands. Two of these sites, the Max-Planck-Institute in Germany, and the University of New Hampshire in the USA, have received funding from national sponsoring agencies to support Guest Investigators at their respective institutions. It is expected that most Guest Investigator activities will take place at these two sites. This does not preclude the possibility, however, of pursuing guest investigations at either the Laboratory for Space Research in Utrecht, or the Space Science Department of ESA/ESTEC in Noordwijk, if appropriate for a particular scientific program.

1. COMPTEL Team Facilities for Data Processing and Analysis by Guest Investigators:

All four sites of the COMPTEL collaboration maintain extensive computing facilities for the analysis of COMPTEL data. The COMPASS data analysis software package has been ported to UNIX platforms at all sites, and is usually run on a network of workstations linked to the master ORACLE database and a mass-storage data archive. A variety of graphical-display and higher-level analysis tools are typically also available. Further, all COMPTEL sites maintain high-speed connections to the Internet and the World Wide Web. The various COMPTEL sites can normally accommodate 2-3 short-term visitors for guest investigations.

2. Data Processing and Analysis by COMPTEL Guest Investigators at a COMPTEL Institution

As described in previous sections, the processing and analysis of COMPTEL data is undertaken with a customized modular software package, COMPASS ("COMPTEL Processing and Analysis Software System"), based on the ORACLE database management system. Specialized tasks within COMPASS perform all routine COMPTEL data processing and analysis. COMPASS is currently installed at each of the four collaborating institutions and is the primary means of access to COMPTEL data. The "production" processing of raw data is carried out at the Max-Planck- Institute in Germany. The low-level processed data is then distributed to other COMPTEL sites for further scientific analysis. Processing runs are monitored automatically by COMPASS at each site, and all sites receive periodic updates of processing activities elsewhere by the routine exchange of a database of "dataset descriptors." Bulk data are transferred between the four sites as requested.

COMPTEL Guest Investigators resident at a COMPTEL institution would be encouraged to become conversant in the use of the COMPASS analysis package, and associated software tools. Given the modular nature of the COMPASS system, it is in principle possible to link special- purpose analysis software provided by Guest Investigators to run within COMPASS; analysis could then be carried out experimentally by the Guest Investigator in a special "domain" available within COMPASS for that purpose. In practice, however, this is a difficult and time-consuming endeavor; the COMPASS manager at the affected site would have to assess the feasibility of such an action on a case-by-case basis. The highest computational priority at any site will always be reserved for the routine processing and analysis of incoming COMPTEL flight data.

At the highest level of processed data (e.g., skymaps) COMPTEL datasets conform as much as possible to standard FITS formats. Such images can be displayed on workstations supporting one of the standard astronomical image processing packages (e.g., IRAF, AIPS, MIDAS) where they can be compared to other data of interest to the Guest Investigator. All COMPTEL sites also share a library of customized IDL-based image-processing software.

COMPTEL data processed by or for a Guest Investigator would most likely be delivered on magnetic tape or, if of sufficiently small size, transferred electronically to the Investigator's home institution. Data delivery will be monitored by the Compton Science Support Center.

Standard data products requested for an approved guest investigation are expected to be available for delivery to the Guest Investigator within 2-4 months of the arrival of the relevant flight data tapes for routine processing at MPE in Germany.

3. Data Processing and Analysis by COMPTEL Guest Investigators at a "Remote" Site

Guest Investigators who have previously visited a COMPTEL institution, or who only wish to receive the highest level of processed data from COMPTEL, may choose to pursue their guest investigation independently at their home institution. It is expected that any use of low-level COMPTEL data at a remote site would require first an extended visit to a COMPTEL institution for an in-depth introduction to the intricacies of COMPTEL data processing.

Processed COMPTEL data for an approved guest investigation would be delivered to the Guest Investigator's home institution on magnetic tape, or sent via electronic transfer. It is not intended at present to export routinely any part of the COMPASS analysis software to individual users. The COMPASS software package is however, installed at the CGRO-SSC where it is available for general GI use via guest accounts

Standard data products requested for an approved guest investigation are expected to be available for delivery to the Guest Investigator within 2-4 months of the arrival of the relevant flight data tapes for routine processing at MPE in Germany.

4. Points of Contact for Further Information

For further information respondents to this Research Announcement may contact the CGRO-SSC. Additional information is available through the World Wide Web pages of the COMPTEL collaboration at <http://wwwgro.unh.edu/comptel/>.

IV. EGRET GUEST INVESTIGATOR PROGRAM

A. Introduction and Scientific Objectives

The broad objective of the Energetic Gamma Ray Experiment Telescope (EGRET) is to make a major advance in high energy (20 MeV to about 30 GeV) gamma-ray astrophysics using a gamma-ray telescope with more than an order of magnitude greater sensitivity and better angular and energy resolution than instruments previously flown. The study of the gamma-ray sky reveals the sites of the most energetic interactions occurring in astrophysics. Because these interactions are generally associated with the dynamic, non-thermal processes in nature, gamma-ray astrophysics provides an excellent opportunity to learn about the forces of change occurring throughout the universe. In addition, since high energy gamma-rays have a very low interaction cross section, they have a very high penetrating power and can reach the Earth from essentially any part of the Galaxy or universe.

Beginning with compact objects and proceeding to the Galaxy, other galaxies, and the universe, the specific objectives of this experiment are:

(1) To search for localized gamma-ray sources in the energy range from 20 MeV to 30 GeV and to measure the intensity, energy spectrum, position, and possible time variations of each;

(2) To improve the knowledge of the locations and nature of known high energy gamma-ray sources not identified with objects seen at other frequencies;

(3) To examine supernova remnants and search for evidence of cosmic-ray particle acceleration revealed by high energy gamma-rays, and, if there are cosmic rays at detectable levels, to study their expansion;

(4) To search for high energy gamma-ray bursts, to study the spectra of low energy gamma-ray bursts, and to contribute to the observation of gamma-rays from solar flares;

(5) To obtain a detailed picture of the diffuse high energy gamma-ray emission from the Galaxy, which in turn will provide information of fundamental importance in the study of the dynamics of the Galaxy (This study should provide a high contrast picture of Galactic structure, including spiral arm segments, clouds, the Galactic center, and the extended disk);

(6) To determine the relative importance of cosmic ray electrons and cosmic ray nucleons throughout the Galaxy by studying the gamma-ray spectrum above 20 MeV and, by combining the results with the continuum radio data, to obtain a better understanding of the Galactic magnetic fields;

(7) To detect and examine the high energy gamma-ray emission from other galaxies, including both normal galaxies and active galaxies; and

(8) To study the diffuse celestial radiation in the energy range above 20 MeV including its energy spectrum and the degree of uniformity, both on small and broad scales (These measurements are of particular importance in relation to cosmological models.).

B. The EGRET Instrument

EGRET is designed to cover the energy range from 20 MeV to about 30 GeV. The instrument uses a multilevel thin-plate spark chamber system to detect gamma rays by the electron-positron pair production process. A calorimeter using NaI(Tl) is placed beneath the instrument to provide good energy resolution over a wide dynamic range. The instrument is covered by a plastic scintillator anticoincidence dome to prevent triggering on events not associated with gamma-rays. The combination of high energies and good spatial resolution in this instrument provides the best source positions of any Compton GRO instrument.

The telescope is shown schematically in Figure IV-1. A gamma-ray entering the telescope within the acceptance angle has a reasonable probability (about 35% above 200 MeV) of converting into an electron-positron pair in one of the thin plates between the spark chambers in the upper portion of the telescope. If at least one particle of the pair is detected by the directional time-of-flight coincidence system as a downward moving particle, and if there is no signal in the large anticoincidence scintillator surrounding the upper portion of the telescope, the track imaging system is triggered, providing a digital picture of the gamma-ray event, and the analysis of the energy signal from the NaI(Tl) detector is initiated. Incident charged particles are rejected by the anticoincidence dome. Low energy backward-moving charged particles which do not reach the anticoincidence dome are rejected by the time-of-flight measurement. Events other than the desired gamma-rays, such as those gamma-rays interacting in the thin pressure vessel inside the anticoincidence scintillator, are rejected in the subsequent data analysis.

The directional telescope consists of two levels of a four by four scintillator array with selected elements of each array in a time-of-flight coincidence. The upper spark chamber assembly consists of 28 spark chamber modules interleaved with twenty-seven 0.02 radiation length plates in which the gamma ray may convert into an electron pair. The initial direction is usually determined from the upper spark chamber data. The lower spark chamber assembly, between the two time-of-flight scintillator planes, allows the electron trajectories to be followed, provides further information on the division of energy between the electrons, permits seeing the separation of the two electrons for very high energy gamma-rays, and shows the entry points of the electrons into the NaI detector.

The energy of the gamma-ray is determined in large part from measurements made in an eight radiation-length thick, 76 cm x 76 cm square NaI(Tl) scintillator crystal below the lower time-of-flight scintillator plane. Spark chamber measurements of Coulomb scattering in the thin plates and position information in the spark chamber system also aid in the energy determination. The energy resolution of the experiment is about 20% (FWHM) over the central part of the energy range. The resolution is degraded to about 25% above several GeV due to incomplete absorption in the NaI calorimeter, and at energies below about 100 MeV where ionization losses in the spark chamber plates comprise an appreciable portion of the total energy. In addition, some particles may completely miss the calorimeter. These losses can be partially recovered through an analysis of the scattering characteristics.

The anticoincidence dome event rate is determined every 1/4 second. These data may be used in the study of gamma-ray bursts and solar flares. Also, the large NaI(Tl) crystal may be used for spectroscopy between 0.6 and 140 MeV independently of the spark chamber system. If a BATSE trigger signal is received, spectra are recorded in four commandable time intervals (normally 0 to 1 sec, 1 to 3 seconds, 3 to 7 seconds, and 7 to 23 seconds) and then sent in the EGRET telemetry stream over the next 35 minutes. In addition, a spectrum between 0.6 and 140 MeV is accumulated every 33 seconds. These spectra are useful for the study of gamma-ray bursts with no associated BATSE trigger, (e.g. when BATSE is reading out a previous event) and for the study of the high-energy portion of the solar flare gamma-ray spectrum. A model of the spacecraft mass and its effect on the gamma-ray spectrum for a specific direction will be available for these studies.

The EGRET operating mode may be altered in two ways. In the common mode, an energy deposition of 7 MeV in the NaI(Tl) calorimeter is required for an event. This mode increases instrument live time by reducing the rate of triggers. It does reduce the sensitivity somewhat (see Table IV-1). For an observation where low energy sensitivity is desired in spite of the poor angular accuracy, it would be reasonable to remove the calorimeter coincidence requirement. The event rate may be reduced by limiting the time-of-flight telescope tile combinations which cause a trigger. The 16 tiles in each of two planes form 96 recognized sub-telescopes which point in 9 different directions, each approximately 0.1 steradians in size. These 9 direction groupings are utilized to limit triggers due to earth albedo gamma-rays, but simultaneously maximize celestial exposure. It would be possible to observe with a subset of these direction modes. Vertical only would give the best efficiency. The field of view would be greatly reduced and not available to others.

The principal characteristics of the EGRET instrument are summarized in Section C. The Co-Principal Investigators of EGRET are Dr. Carl E. Fichtel of the NASA/Goddard Space Flight Center (GSFC) and Dr. Klaus Pinkau of the Max Planck Institute for Plasma Physics. Co-Investigators are at the Goddard Space Flight Center, Stanford University, the Max Planck Institute for Extraterrestrial Physics, Grumman Aerospace Corporation and Hampden-Sydney College.

The EGRET Instrument

Figure IV-1. The EGRET Instrument

C. Instrument Parameters and Capabilities

1. Type: spark chambers, NaI(Tl) crystals, and plastic scintillators.

2. Energy Range: 20 MeV to about 30 GeV

3. Energy Resolution: approximately twenty percent over the central part of the energy range.

4. Total Detector Area: approximately 6400 cm2

5. Effective Are : approximately 1500 cm2 between 200 MeV and 1000 MeV, falling at higher and lower energies.

6. Point Source Sensitivity: varies with the spectrum and location of the source and the observing time. Under optimum conditions, well off the galactic plane, it should be approximately 6 x 10-8 cm-2s-1 for E > 100 MeV for a full two week exposure. See Section VI of this Appendix for more detailed sensitivity information.
7. Source Position Location: Varies with the nature of the source intensity, location, and energy spectrum from 5 - 30 arcmin.

8. Field of View: approximately a gaussian shape with a half width at half maximum of about 20. Note that the full field of view will not generally be used.

9. Timing Accuracy: 0.1 ms absolute

10. Weight: about 1830 kg (4035 lbs)

11. Size: 2.25 m x 1.65 m diameter

12. Power: 190 W (including heater power)

D. Data Reduction and Analysis

The EGRET instrument is a wide field telescope even when only one of its subtelescope directional modes is used. Each observed gamma ray is detected individually. Even though the instrument is designed to detect high energy gamma rays, much less than half of the events resulting from spark chamber triggers satisfy the rigid rules for finally accepted celestial gamma ray events. It is the EGRET Science Team's responsibility, together with their EGRET team of data analysts and programmers, to separate the gamma rays from other triggers and determine within errors the direction and energy of each gamma ray based on carefully quality controlled procedures and the calibration data. An entire observation is analyzed as a set since there is no way to select gamma rays or subsets prior to analysis. The direction will be given in terms of galactic, celestial and Earth coordinates. It is the responsibility of the EGRET Science Team to make a determination of the diffuse celestial radiation, galactic and extragalactic, to the best of their ability for all regions observed. They are also responsible for determining source strengths and spectra in a consistent, carefully reviewed fashion and maintaining a catalog. They will also provide upper limits, when it seems appropriate, for a potential source. When a Guest Observer source has been named and the data from the appropriate observing period analyzed, he or she will be given the information on the relevant photons and the estimated source strength and spectrum or upper limit. These data will be withheld from the public domain including the catalog for the duration of the proprietary period after it is given to the Guest Observer.

The telemetered data from EGRET are organized into packets of 1,786 bytes that span a time interval of 2.048 seconds. Each packet contains a header with the time of day to the nearest second, and spacecraft position, pointing, and earth-center coordinate information. Following the header, housekeeping data consisting, for example, of voltages, currents, pressures, temperatures, command verification, counter rates, and instrument operating modes are included. Spectral data from the Total Absorption Shower (~ 0.6 to ~140 MeV) Counter (TASC) are routinely sent with each set of 16 packets. Likewise, burst data, when present, are encoded into a series of 1,024 packets. The remainder of the packet data consists of individual event information. Events are associated with a triggering of the spark chamber and consist primarily of coordinate readouts of chamber tracks, although other information, including a precise time vernier, the set of sub-telescopes that were triggered, and the TASC energy measurement are part of each event data. Since gamma rays can trigger the detector at any time, the event data are asynchronous to the packet formation. Events are therefore stored in a buffer on-board and are moved from there to the packets during their assembly by the on-board firmware. Up to 12 events of variable length may occur within any packet, and events may be continued from one packet to the next.

Production processing of the telemetry information consists of decoding the packet data, scaling all housekeeping data to engineering units, assembling the TASC spectra from the appropriate set of packets, separating and assembling events along with event related information, and, finally, analyzing each event. During the course of one day, approximately 7.5 Megabytes of housekeeping are accumulated to monitor instrument performance and to determine the sky exposure to gamma-rays. The TASC will generate 1,760 spectra, and the spark chamber is triggered approximately 5 x 104 times.

The event analysis is the most critical part of the entire ground analysis system. Each event contains spark coordinate information in two orthogonal views. First, it is necessary to structure the event. This process involves determining which spark coordinates to associate with tracks from the gamma-ray interaction, and then to correlate the multiple track projections in the two views of the read-out. The usual signature of a gamma-ray interaction is an inverted "V" formed by the electron and positron that issue from the point where the gamma-ray conversion occurs.

Once the tracks are properly structured, the event is screened to assure that the appropriate signature has been found and that the interaction occurred within the spark chamber volume. A multiple-scattering analysis is then performed for each track to estimate the energy of each secondary particle. Analysis of the total energy of the event is made using the data from the TASC and correcting for energy loss in the plates between the spark chambers, and for leakage of particles out the side of the detector, using the track information and the scattering results. When both the scattering and energy calculations have been made, a procedure is followed to select the best energy values for the two principal tracks that form the vertex. A least-squares fit to the points of each track in each view is made where the number of points used is based on track straightness. A weighted bisector of the angle between the tracks is constructed for each orthogonal view. This direction, together with the pointing direction of the instrument, is used to determine the arrival direction of the gamma-ray. In summary, the event processing provides arrival time, energy, and direction for each gamma-ray.

The computer analysis that generates the initial structuring must be closely monitored by trained personnel and knowledgeable scientists who study the two orthogonal views of analyzed events in light of the developed rules and comparisons to calibration data. If improvements are to be made in the identification of the sparks associated with tracks, an analyst can make these changes using an interactive graphics workstation. The production processing system assigns categories of potential structuring problems which are then selectively reviewed manually. In addition, a random selection of events are also reviewed to insure that the system is meeting standards.

The most time consuming task, requiring trained scientists and data analysts, is the individual gamma-ray event studies on the graphics units for those events which have been flagged by the automatic analysis as needing more study for any of several specific reasons. Some of these include: uncertainty in the automatic program regarding whether the event comes from the wall, inability to satisfactorily match the two orthogonal views for any of several reasons, or starting point difficulties. This tedious process must be accomplished for an entire field of view before any of the results from that field of view can be determined since not only gamma-ray identification, but also direction and energy may be affected. The series of steps outlined above must be performed for all photons in the field of view for a given exposure before the sky map can be generated and the point source analysis be done.

The determination of the exposure factors is especially complex for EGRET. The telescope is divided into 96 sub-telescopes that are organized into groups that point in 9 different directions. During any orbit, the directional groups are dynamically commanded "off" and "on" by the Compton GRO computer depending on whether or not the Earth is in the field of view for that direction. The 9 sets of direction groupings are not independent. That is, an event may be able to trigger more than one mode. The direction modes therefore must be considered in combinations. A separate set of instrument performance properties of efficiency, angular resolution, and energy resolution as a function of energy must be referenced for each of 74 combinations in order to construct the exposure. If the TASC threshold or instrument triggering requirements are changed, another set of tables must be generated for use. Typically the determination of exposure will involve integrating the product of live-time, detection efficiency, and solid angle for each bin on the sky, taking into account the time varying mode changes. Any hardware anomalies that require turning off part of the telescope system will require that complete new performance tables be generated from the calibration event data by eliminating events that could not have been detected if the disabled portion of the instrument had not been active. A software system will be in place to generate exposures from the housekeeping data of the instrument, but it must be carefully monitored, especially if trigger requirements or thresholds are changed, or if any of the coincidence tubes have to be turned off.

The combination of overlapping fields of view is desirable in order to increase the statistical significance and, in some cases, to reduce the uncertainties produced by the low sensitivity for an object far from the axis. The analysis of a point source must also take into account the energy dependence of the angular resolution and the background spatial distribution, including its energy dependence.

The result of the standard production processing is a comprehensive EGRET primary database (PDB). From the PDB, a secondary set of databases are constructed which serve as the input for the scientific analysis. The calibration database contains information needed to produce spectra and skymaps and build the event databases. Two other crucial databases are a summary database which is a condensed form of the gamma ray event database and will be the primary input for many scientific analyses and an exposure history database which contains pointing and instrument mode information which is needed for determining the exposure maps.

An extensive library of programs exist for manipulating and analyzing science data. For instance, skymaps can be developed with the programs MAPGEN, INTMAP, and SKYMAP. MAPGEN uses the summary database to create or update gamma-ray skymaps for a user specified range of arrival directions, energy and time ranges in a variety of coordinate systems.. The output of MAPGEN is a FITS format file. The INTMAP module takes the maps created by MAPGEN and utilizes the timeline and exposure history files to create corresponding exposure and intensity maps, again in FITS format. These files can also serve as the input to SKYMAP which is useful for interactively displaying and plotting skymaps. SKYMAP capabilities include rescaling the map, adding contours, and displaying multiple maps. The program SPECTRAL is used for the analysis of the energy spectrum of the detected source.

If the temporal properties of a source are of interest, PULSAR and SEARCH are available to analyze the time series for an observation. PULSAR is an interactive program which will fold the data at an assumed pulsar period and generate the light curve. Proper barycentering, period derivative, and binary orbit corrections are made if needed. For searches over a range of trial frequencies, SEARCH will allow the user to apply statistical tests to the set of arrival times over a user specified range of periods. The significance as a function of frequency is the most important output of SEARCH.

Other programs and tables will be available in the near future. Theses include the numerous tables in the EGRET Phase 1 catalog, a program to give source positions and strengths, a table of estimates of the gamma-ray diffuse radiation, and a program to search for pulsed radiation within a limited range of parameters.

E. EGRET Guest Investigator Opportunities

For Cycle-6 and beyond it is especially important to read Section I. of this chapter. The Guest Investigator Program provides for both collaborative and independent modes of participation. In the collaborative mode, the Principal Guest Investigator would choose to work closely with the EGRET scientific team. Note however that team resources in support of such activity are much more limited than in the earlier stages of the mission. Anyone interested such a collaboration should check the appropriate boxes in the "Proposal Type Form" in appendix D of the NRA.

In the independent mode the Guest Investigator will receive support from the Compton Science Support Center.

Some investigations such as the diffuse Galactic and extragalactic emission, the search for and analysis of high energy microsecond gamma ray bursts, and the development of source catalogs are not included in the NASA Research Announcement. The nature and especially the extent and degree of effort involved in these observations and analyses make them inappropriate for guest observers.

Like all experimenters on NASA spacecraft, Compton GRO Instrument Team and Guest Investigators are required to provide a processed and accessible version of data, along with instructions for use, to the National Space Science Data Center and/or other NASA managed public data archive after the proprietary period expires. These data are available upon request to anyone. The Compton Science Support Center provides information on data availability, assistance in retrieving the data, instrument properties and response, and some analysis software.

Guest Investigators visiting the Goddard Space Flight Center will have access to a variety of hardware and software tools available to supplement their own resources. Additionally, most of the EGRET analysis software is available from the CGRO-SSC for installation at ones home institution.

F. Feasibility Evaluation for EGRET Observations

In determining the feasibility of an EGRET observation, it should first be determined whether or not the Compton GRO all-sky survey has obtained the desired data already. The EGRET all-sky survey exposure is typically 6 x 108 cm2s for the central part of the energy range, but is not uniform. The Compton Science Support Center has an on-line facility(1) which may be used to obtain the actual and planned exposure for any specific direction. The proposer should state why additional EGRET observing time, beyond that from the all-sky survey, is being requested. Reasons for new observations could include: variability or suspected variability of the source; requirement for longer exposure or additional exposure time beyond that provided on the source by the all-sky survey; and the desirability of simultaneous observations with other instruments covering different spectral ranges.

Typical observing periods will normally be the equivalent of 2 weeks with full telemetry coverage; so a discussion should be presented showing that the proposed observation can reasonably be expected to succeed given one or more of these standard observing times. If the calculations show that the observation is unlikely to be successful, but the proposer wishes to request it nevertheless, then some justification for this request should be provided (for example: the target is close to an already planned target so no additional observing time is needed; or the science is sufficiently compelling and the target brightness sufficiently uncertain that it is worthwhile to attempt a very uncertain detection).

The value quoted above for EGRET sensitivity, of 6x10-8 photons cm-2s-1, refers in general to a 3- detection of a point source located well outside the galactic plane and optimally positioned in the EGRET field-of-view observed for 2 weeks of time. Other considerations apply for other scientific objectives, and must be taken into account in determining the actual feasibility of a proposed observation. See Section VI of this Appendix for more detail on the EGRET sensitivity. For EGRET, T = 0.40+/-0.05 is the observing efficiency (due to earth occultation, instrument dead time, and SAA passage) for the observation time T. (Note: the operation of the Compton Observatory without the tape recorders significantly reduces T below the factor quoted, by a factor which depends on several variables.) It is suggested that the slight azimuthal dependence in sensitivity be ignored and that the effective area be treated as a separable function, i.e.

A(E,,) = AE(E) A()

Table IV-1 gives the approximate values of AE(E) while Figure IV-2 gives data for A()

Energy

(MeV)

AE(TASC not

in coin.)

AE (TASC

in coin.)

35 0.3 x 103 cm2 0.07 x 103 cm2
100 1.1 x 103 0.9 x 103
200 1.5 x 103 1.4 x 103
500 1.6 x 103 1.5 x 103
3000 1.1 x 103 1.1 x 103
10,000 0.7 x 103 0.7 x 103

Table IV-1. Preliminary EGRET effective area (on axis).

The EGRET sky-survey was completed with TASC in coincidence in order to reduce spark chamber triggers and to improve instrument live time. Unless the enhanced effective area available with TASC out of coincidence can be shown to offset these disadvantages, TASC should be in coincidence.

For purposes of the Phase 4 proposals, the EGRET point spread function (psf) may be considered to be a gaussian, independent of angle with respect to the instrument axis, with preliminary rms values as a function of energy given in Table IV-2. These data will be superseded by more accurate functions to be made available to the guest investigator for the actual data analysis.

Energy

(MeV)

psf

(deg)

35 < E < 70 4.3
70 < E < 150 2.6
150 < E < 500 1.4
500 < E < 2000 0.7
2000 < E < 30000 0.4

Table IV-2. Preliminary EGRET point spread function (projected rms values).

To determine the background flux, the COS-B and SAS-2 results can be used. For galactic latitude |b| < 20, the COS-B results(2) have much better statistics than the SAS-2 results(3), but the instrument background is much larger. At high galactic latitude, only the SAS-2 results(4) are useful because of the COS-B instrumental background. The Compton Science Support Center plans to provide a facility which the proposer might use remotely to determine the diffuse background in any direction and at any energy 30 < E < 20,000 MeV.

For the purpose of proposing spectral analysis, the full width at half maximum for the energy dispersion (dE/E) should be considered to be 20% for 200 < E < 3000 MeV, and 25% for 60 < E < 200 or 1000 < E < 30,000 MeV. The detailed EGRET energy dispersion data will be made available to the guest investigators for the actual data analysis.

G. Specification of EGRET Observations

The Compton GRO Observation Definition Form in Appendix D should be submitted by all proposers who request data from EGRET observations of specific targets (Proposal Types 1 through 4). The EGRET instrument has well-defined data types available and there are relatively few instrument configurations which can be selected. This makes defining an observation reasonably simple. The quantities which must be specified include

1. Target name and pointing for the instrument. Note here that it is not always the case that EGRET observations will have the target centered. If multiple targets are to be observed simultaneously, a pointing for the center of the FOV should be specified so as to allow all targets to be observed. Equinox 2000 celestial and/or galactic coordinates should be used.

2. Duration of the observation. This should be stated with reference to the discussion of feasibility covered in section F. The time to be entered is the net observing time (i.e. the time during which the target is in view, and the spacecraft is not in the SAA, and telemetry is being received on the ground, so that photons can be collected). At scheduling time, the observation duration may be adjusted to compensate for the loss in sensitivity for off-axis observations shown in Figure IV-2, and for the Compton GRO commanding cycle of days or weeks.

The effective exposure time at launch, with full telemetry coverage, was actually about 0.4 of the number of days of observing due to target occultation, SAA periods, and dead time. At present the efficiency is somewhat less. For example, the net exposure time for a 2-week observing period for an on-axis target with full telemetry coverage would be 0.8 weeks. 3. Non-standard instrument modes. Proposers should note that such modes may require special calibration data or have other difficulties and therefore require special justification.

Inclination Angle Dependence of EGRET Sensitivity

Fiigure IV-2

4. Any special requirements for correlative observations or other reasons, such as critical observing times or time separations for variable sources, or for simultaneous observations by one or more other Compton GRO instruments or other observatories.

5. Expected data products (optional). Use the Compton GRO Data Products Definition Form. See section H for a description of some standard products. Anticipated requests for non-standard products should be included.

H. Data Products to be Provided to EGRET Guest Investigators

The type of data products received will depend upon the requirements of the Guest Investigator both in terms of the requested observation and the resources available to the investigator, however, the intensity or upper limit for the proposed source will always be provided. The data can be distributed both through network retrieval and by tape media.

The production time for most data products is expected to be 3 to 4 months with receipt by the guest investigator following soon after. A large part of this time is for the manual editing and inspection of candidate gamma-ray events.

Documentation will be available through the Compton Science Support Center. This includes descriptions and users guides for data analysis software, information on instrument configuration and specifications, and information regarding the science previously and currently being done with EGRET. Contact the Compton SSC for more information on documentation.

I. Use of EGRET During Cycle 7 and Beyond

Beginning in Cycle 6, an alternate procedure has been followed in order to extend the life of EGRET, wherein observations using only the central telescopes, or at most a subset of the total sub-telescopes, are being encouraged. For any viewing period, one of the following choices is available:

1. Use only the on-axis telescopes to look at a single source or a few sources close together. The approximate solid angle is shown in Figure IV-3. The gas usage is about one-third of that when all modes are activated.

2. Use the on-axis telescopes plus the two adjacent sets in the same plane. (This mode could, for example, be used to study a region of the galactic plane.) See the caption for Figure IV-3 for an indication of the solid angle. The gas usage is just over one-half of that when all modes are activated.

3. Use some other telescope combination, such as the on-axis telescopes plus one adjacent set. (There would probably be a small analysis delay since special software runs for the chosen mode would be required.) See Bertsch et al. (1989) for a description of the direction modes.

The Solid Angle for EGRET

Figure IV-3.
The Solid Angle for EGRET. The solid angle is a complex three-dimensional function of energy; however, the enclosed figure combined with the efficiency curve as a function of energy gives a reasonably good understanding. In this figure, the upper curves for 100 MeV and 1 GeV give the solid angle for the standard mode with all telescopes activated. The lower curves give the solid angle for the same energies with only the central telescopes active. In the "fan" mode described in section IV-I, the solid angle at 1 GeV is approximately represented in one plane by the upper curve and in the orthogonal plane by the lower curve. At 100 MeV, a similar statement is approximately correct except the true curves are somewhat below the upper curve and somewhat above the lower curve. Detailed sensitivities for the standard energy internals will be provided by the EGRET team.4. Use the full compliment of telescopes. This mode will be used sparingly and only in the case of exceptionally strong scientific justification.

5. Leave the EGRET spark chamber system turned off if the scientific return from a given viewing direction is expected to be low. All other systems remain in their present configuration. The spark chamber system which would only be activated as indicated in the last paragraph or as indicated below. The spark chamber and trigger system can be fully activated relatively quickly (approximately 2 seconds) upon receipt from BATSE of a high intensity burst in the field of view.

Observations requiring the use of EGRET spark chamber system during the remainder of the mission will be allocated on the basis of the scientific merit of a proposal (as determined by peer review), also taking into account the limited resource of the remaining gas supply. It is anticipated that the rate of usage realized will be about 1/3 of the usage rates over the mission prior to Cycle 6.

Some reserve spark chamber system capability will be kept as late in the mission as possible for the observation of a great solar flare or an object of special interest to obtain the greatest possible time differential.

References

Bertsch, D.L., et al. 1989, Proc. Gamma-Ray Observatory Science Workshop, Ed. W.N. Johnson (Greenbelt, NASA), 2-52
Thompson, D.J., et al. 1993, ApJ Supp. 86, 629-656

1. This facility, and various other utilities and information are available on the CGRO WWW site at <http://cossc.gsfc.nasa.gov/>.

2. Mayer-Hasselwander, H., et al., 1982, Astron. Astrophys, 105,167.

3. Hartman, R.C., et al., 1979, Ap.J., 230, 597

4. Fichtel, C.E., et al., 1977, Ap.J., 217, L9

V. BATSE GUEST INVESTIGATOR PROGRAM

A. Introduction and Scientific Objectives

BATSE (the Burst And Transient Source Experiment) consists of eight uncollimated detector modules arranged on the corners of the Compton Gamma Ray Observatory to provide the maximum unobstructed view of the celestial sphere. Each detector module contains a Large-Area Detector (LAD), optimized for sensitivity and directional response, and a spectroscopy detector (SD) optimized for broad energy coverage and energy resolution. A description of the instrument is given below in section B. Further information is available in several papers in the First GRO Science Workshop (Fishman et al. 1989; Paciesas et al. 1989) and in a NASA report (Horack 1991). The basic parameters and capabilities of the BATSE instrument are summarized in Table V-1.

Table V-1.

BATSE Instrument Characteristics

Detectors
Number of Detectors Eight Large Area Detectors (LAD)
Eight Spectroscopy Detectors (SD)
Field of View Full Sky
Sensitive Area 2025 cm2 per LAD
127 cm2 per SD
Energy Range 20 keV - 1.9 MeV for LAD
10 keV - 100 MeV for SD
Experiment Sensitivities
Burst Sensitivity 3 x 10-8 ergs/cm2 (1 sec burst)
Time Resolution 2 s minimum
Burst Location Accuracy 4.0 (strong burst)
Occultation Sensitivity (3 ) 100 mCrab (30-100 keV, one day)
FOB1 Pulsed Source Sensitivity (3 ) 0.12 pulsed Crab (30-250 keV, typical day)
FOG2 Pulsed Source Sensitivity (3 ) 0.05 pulsed Crab (30-250 keV, typical day)

1 Folded-On-Board with special hardware 2 Folded-On-Ground

The primary objective of BATSE is the detection, location and study of gamma-ray bursts. With its large detecting elements and versatile data system, BATSE is providing the most sensitive and comprehensive set of observations of gamma-ray bursts yet obtained.

All eight detector modules have been operating well since the experiment was activated on April 21, 1991. Since activation, the rate of gamma-ray bursts observed by BATSE has been approximately 0.85 per day. Correcting for Earth blockage, experiment dead-time due to false triggers and for operational reasons (primarily SAA passages), the full-sky burst rate at the BATSE sensitivity is estimated to be about 800 bursts per year.

Directions to burst sources are determined by comparing count rates among separate detectors. Significant corrections are needed to account for the effects of atmospheric scattering. At present, the burst location accuracy is estimated to be approximately 4, being limited by systematic errors. This estimate is based on the computed location of solar flares and from gamma-ray bursts located by the current interplanetary burst network.

The most significant and unexpected observation made by BATSE thus far is the isotropy of gamma-ray bursts. This observation coupled with statistical evidence for a non-homogeneous spatial distribution has ruled out the galactic disk distribution expected by many before launch. The observed time profiles of individual gamma-ray bursts show an enormous diversity in overall shape, pulse structure and duration. Models of bursts must be able to account for this diversity.

The Earth occultation technique is being used to detect transient sources above 20 keV and to study variability in the stronger sources that are detectable by the BATSE LADs. Virtually every source that BATSE has observed is variable to some degree, the notable exception being the Crab.

Many pulsars can be observed by BATSE using the distinct temporal characteristics of the sources. This is accomplished by epoch folding on-board the experiment or on the ground, or by fast Fourier transform analysis on the ground. So far BATSE has detected 19 pulsars.

The high sensitivity and time resolution of the BATSE detectors at energies above 20 keV, coupled with the large duty cycle for solar viewing, provide BATSE with some unique capabilities. To date BATSE has triggered on more than 700 flares and detected more than 2000.

B. The BATSE Instrument

One of the eight identically configured detector modules of BATSE is shown schematically in Figure V-1. Each detector module contains two NaI(Tl) scintillation detectors: a Large Area Detector (LAD) optimized for sensitivity and directional response, and a Spectroscopy Detector (SD) optimized for energy coverage and energy resolution. The eight planes of the LADS are parallel to the eight faces of a regular octahedron, with the orthogonal primary axes of the octahedron aligned with the coordinate axes of the Compton spacecraft.

The LAD detector is a disk of NaI scintillation crystal 20 inches in diameter and one-half inch thick, mounted on a three-quarter inch layer of quartz. The large diameter-to-thickness ratio of the scintillation crystal produces a detector angular response similar to that of a cosine function at low energies where the crystal is opaque to incident radiation. At energies above 300 keV, the angular response is flatter than a cosine. A light collector housing on each detector brings the scintillation light into three 5-inch diameter photo-multiplier tubes (PMTs). The signals from the three tubes are summed at the detector. A quarter-inch plastic scintillation detector in front of the LAD is used as an anticoincidence shield to reduce the background due to charged particles. A thin lead and tin shield inside the light collector housing reduces the amount of background and scattered radiation entering the back side.

The spectroscopy detector is an uncollimated NaI(Tl) scintillation detector 5 inches in diameter and 3 inches thick. A single 5 inch photo-multiplier tube is directly coupled to the scintillation detector window. The housing of the PMT has a passive lead/tin shield similar to that of the LADs. The crystal housing has a 3-inch diameter 50 mils thickness beryllium window on its front face in order to provide high efficiency down to 10 keV for exposures near detector normal. The axis of symmetry of an SD is offset from by 19 from the LAD axis for mechanical reasons. This offset is toward the spacecraft x-y plane.

One of the eight identically configured BATSE detector
modules

Figure V-1. One of the eight identically configured BATSE detector modules. Each module contains a Large Area Detector (LAD) and a Spectroscopy Detector (SD).

Scintillation pulses from the detectors are processed by a gated baseline restoration circuit in order to minimize spectral distortion at high counting rates. Pulses are processed in parallel by a high-speed, four-channel discriminator circuit and by a slower pulse height analyzer system. The nominal equivalent energies of the upper three discriminators for the LADs are 60, 110, and 325 keV. The lower-level discriminators are programmable, and currently set near 20 keV. Two of the fast discriminators for the SDs are set at energies above the energies analyzed by the pulse height system. The gain of each detector system is determined by the high voltage applied to the PMTs. The SDs have been operated at different gains in order to span from 10 keV to greater than 100 MeV.

Each of the eight BATSE detector modules sends data to the Central Electronics Unit (CEU). The CEU contains hardware and software that accumulates the data into several large RAM memory buffers. Extensive use of commandable parameters, plus the capability to reprogram the flight software, insures that BATSE will have the flexibility to respond to unforeseen conditions in orbit or newly discovered -ray phenomena. Signals from the pulse-height converters are used to construct 128-channel spectra from the LADs and 256-channel spectra from the SDs. Each of the spectra are subdivided into ranges with different dispersions to increase the dynamic range and to efficiently use the available telemetry space. These energy channels are also mapped into 16 coarse energy channels using programmable lookup tables, one for the LADs and one for the SDs. This permits a tradeoff of time resolution for energy resolution in several of the data types. Discriminator events are accumulated in the hardware every 64 ms. The CEU hardware constructs various data types from the discriminators, 16-channel, 128-channel, and 256-channel spectral data.

The CEU produces data for telemetry at a uniform rate of one packet every 2.048 seconds. Each of these packets contains LAD and SD discriminator rates, 16-channel LAD spectra, and housekeeping information. In addition there is a variable portion of the packet which normally contains either high resolution spectral data or epoch folded data for pulsar observations. The CEU contains hardware for folding 16-channel data on board at the approximate period of a candidate pulsed source from selected detectors. Two pulsars of different periods can be simultaneously acquired, however the same set of detectors must be selected for both. The accumulation of pulsar data and high resolution LAD and SD spectra is driven by an on-board schedule that is synchronized with the Compton orbit. This schedule is updated with changes in Compton spacecraft orientation.

BATSE detects -ray bursts on-board by examining the count rates of each of the eight LADs for statistically significant increases above background on each of three time scales: 64 ms, 256 ms, and 1024 ms. The discriminator rates in channels 2 and 3 (approximately 60 to 325 keV) are used. The background rate is determined for each detector over a commandable time interval currently set at 17.4 seconds. The statistical significance required for a burst trigger is set separately for each of the three time scales, with a quantization of 0.0625s. These thresholds are currently set at 5.5s. At least two detectors must exceed threshold for a burst trigger to occur. An additional requirement for burst triggering is that the detector with the greatest increase in count rate must have an increase in the charged particle rate that is less than a specified fraction of the increase in the neutral rate. This is done in order to avoid triggering on charged-particle event encounters, such as those produced by spacecraft containing nuclear reactor power sources.

When a -ray burst is detected, the CEU enters a fast data acquisition mode and rapidly stores a variety of data types into memory. Over a interval of 45 to 105 minutes (depending on burst intensity) the normally scheduled output of pulsar data and high resolution spectra is suspended, and the collected burst data is read out in the variable portion of the data packets. The normal output schedule then resumes, in proper synchronization with the Compton orbit. While the burst memories are being telemetered, the trigger thresholds are temporarily revised to values corresponding to the maximum rates detected during the burst. Thus, a stronger burst will terminate the readout of a weaker burst and overwrite the burst memories. The data available from a overwritten burst is timing dependent, but includes at a minimum the 64 ms resolution discriminator data (DISCSC).

The failure of the onboard tape recorders has led to a loss of some burst data. New flight software now corrects this problem. With the new software burst readouts are suspended during telemetry gaps. A new data type, DISCLB, is collected to compensate for the loss of DISCSC and DISCLA data during telemetry gaps. In addition, three channels of the DISCLA data are collected during gaps in the normal telemetry, using a 1-kbyte/sec telemetry channel.

Whenever a burst trigger occurs BATSE provides the other Compton instruments a signal that may be used to initiate special burst data processing modes. If the burst source can be observed by OSSE, BATSE also provides an estimate of the required elevation angle.

C. Data Reduction and Analysis

The Daily Data Sets. Every day the BATSE data packets for the previous day are electronically transmitted to MSFC. This day's worth of data is unpacked and sorted into a group of files called a Daily Data Set. After a set of routine analyses are performed upon the Daily Data Set it is archived onto magneto-optical discs.

The data within the Daily Data Set can be grouped into three classes; burst data, scheduled data, and background data. The high time resolution burst data is gathered when a -ray burst, solar flare, or other event meets the BATSE burst trigger condition. The scheduled data is available from orbits during which no burst trigger occurred. It consists of high resolution spectra and data from the on-board folding of pulsars. Its content is flexible and needs to be pre-planned. The background data consists of discriminator rates and medium resolution LAD spectra that are collected continuously. These data-types are summarized in Table V-2 and discussed below.

Burst Data Types

DISCSC. The output of Discriminator Science Data (DISCSC) begins immediately after a burst trigger and continues for the whole burst acquisition. DISCSC consist of 64 ms discriminator data summed from each of the LAD's that were triggered by the burst.

PREB. The Pre-Burst (PREB) data consists of discriminator counts from the eight LADS immediately before a burst trigger. In the normal operating mode the LAD discriminator rates are stored in a ring buffer. This makes available the 64 ms resolution LAD discriminator rates for the 2.048 seconds immediately prior to a burst trigger.

DISCLB. Discriminator data from the eight LAD's collected to compensate for the loss of DISCSC and DISCLA data during telemetry gaps. This consists of 16.384 s of pre-trigger data with 2.048 second resolution and 49.154 s of post-trigger data with 1.024 second resolution.

TTE. Time-Tagged Event (TTE) data provide the time of occurrence, discriminator channel, and detector number of 32,768 individual LAD events. The time resolution is 2 ms. The memory is organized as a continually running ring buffer. When the burst trigger occurs, data are entered only from the LADs that trigger the burst. Accumulation of data halts after three-fourths of the memory fills. The other one-fourth of the memory then contains data from all eight LADs immediately prior to the trigger.

STTE. Spectroscopy Time-tagged Event (STTE) data provide similar data for the SDs with a time resolution of 128 s. There are four memory buffers of 16,384 event capacity each. A selectable subset (currently two) run continually and the others are filled successively if a burst occurs. With the current settings the STTE data available from before the burst trigger can range from 16,384 to 32768 events. Normally the pre-trigger STTE includes all eight detectors, however if a subset of SD detectors were being used for on-board pulsar folding when the burst occurred, only data from this subset are included. The post STTE data are from the triggered detectors.

Table V-II

BATSE Data Types

Data Type Detectors Energy Channels Time Resolution Description
Burst Data
DISCSC 2-4 LADs1 4 64 ms Discriminators after trigger
PREB 8 LADs 4 64 ms Discriminators prior to trigger
DISCLB 8 LADs 4 1.024 s Discriminators
TTE 2-4 LADs2 4 2 ms Time-tagged events
STTE 2-4 SDs3 256 128 ms Time tagged events
TTS 2-4 LADs1 4 2 ms Time to spill data
MER 2-4 LADs1 16 16-64 ms Medium resolution spectra
HERB 4 LADs4 128 64 ms High resolution spectra
SHERB 4 SDs4 256 *128 ms High resolution spectra
Scheduled Data
HER 8 LADs 128 *16.384 s High resolution spectra
SHER 8 SDs 256 *32.768 s High resolution spectra
PSRFUL Any 16 1 ms5 Full readout pulsar data
PSR16 Any 16 1 ms5 Folded on board pulsar data
PSRSEL Any 4 1 ms5 Short readout pulsar data
PSRSUM Any 4 1 ms5 Short readout pulsar data
Background Data
DISCLA 8 LADs 4 1.024 s Discriminators
DISCSP 8 SDs 4 2.048 s Discriminators
CONT 8 LADs 16 2.048 s Medium resolution spectra
  • 1 Summed over the triggered LADs.
  • 2 Pre-burst data is from all modules. Post-burst data is from triggered modules.
  • 3 As in 2 under normal conditions, however see text.
  • 4 From the four modules with highest LAD rates at trigger.
  • 5 Resolution of the folding period. Folded with 64 bins/cycle.

    TTS. The Time-to-Spill (TTS) data provide the time to accumulate a specified number of discriminator events from the LADs. The number of events is nominally 64, but can also be commanded to either 16 or 256. Discriminator events are summed from those detectors that trigger the burst. The time resolution is 1 s for the shortest times, but increases at longer times to allow a satisfactory dynamic range. Each discriminator energy channel operates independently and has a capacity for 16,384 entries.
  • MER. Medium Energy Resolution (MER) data consist of 4,096 16-channel spectra summed from triggered detectors. The nominal accumulation time is 16 ms for 2,048 spectra, followed by 64 ms for the remaining 2,048 spectra for a total duration of about 164 s. These accumulation times are commandable software parameters.

    HERB. The High Energy Resolution Burst (HERB) data consist of 128 separate HER spectra stored during a burst. The selection of detectors and storage times is under software control. Nominally, the accumulation time will be rate-dependent. Spectra from four LADs will be stored, with the more brightly illuminated detectors stored more often.

    SHERB. The Spectroscopy High Energy Resolution Burst (SHERB) data provide the same function for the SDs, the only difference being that a total of 192 256-channel spectra can be stored.

    Scheduled Data Types

    HER. High Energy Resolution (HER) data consist of 128-channel spectra from each of the eight LADs. These data are output in eight consecutive packets, so the minimum time resolution is 16.384 s.

    SHER. Spectroscopy High Energy Resolution (SHER) data consist of 256-channel spectra from each of the eight SDs. A sequence of 16 packets is used to transmit the SHER data, so that the minimum time resolution is 32.768 s.

    PSRFUL. The CEU hardware folds 16-channel data on-board at the approximate period of a candidate pulsed source. The folding period is specified as an integer number of ms. The data which is folded may be selected as either the sum over a set of LADs or the sum over a set of SDs. The hardware folds this data into 64 phase bins over a commandable number of folding periods. Data for two pulsars may be acquired simultaneously providing that the same selection of detectors is used. The Full Pulsar (PSRFUL) data contains the complete on-board accumulation. The 1024 (64 phase bins by 16 energy channels) 12 bit words are read out in a sequence of 8 packets.

    PSR16. The remaining pulsar data types represent different ways of compressing the data for output. For the 16-channel Pulsar (PSR16) data the full accumulation is read out with single byte precision in a sequence of 4 packets. Overflow information is transmitted which normally allows reconstruction of the rates in each phase bin.

    PSRSEL. For the energy Selected Pulsar (PSRSEL) data the 64 phase bins for four sequential energy channels is read out with single byte precision in a single packet.

    PSRSUM. For the energy Summed Pulsar (PSRSUM) data the 16 energy channels are summed into four. The 64 phase bins for these four summed energy channels are then read out with single byte precision in a single packet.

    Background Data Types

    DISCLA. Large Area detector Discriminator (DISCLA) data consists of 1.024 sec accumulations of the four discriminator channels, the total counts (including charged particle detector coincident events), and the charged particle detector counts for the eight LADs.

    DISCSP. Spectroscopy detector Discriminator (DISCSP) data consists of 2.048 sec accumulations of the four discriminator channels for the 8 SDs.

    CONT. The Continuous (CONT) data consists of 2.048 second resolution 16-channel count spectra from the eight LADs.

    Higher Level Processing

    Preparation of the daily datasets is the responsibility of the BATSE mission operations team. This team also generates calibration files, maintains several databases and performs standard analyses for science teams. The science teams analyze the data and supervise the production of higher level data products for guest investigators and archival delivery. The available higher level products are discussed in section H.

    Burst analysis is the primary BATSE science objective. All bursts that trigger the on-board data system are processed into a set of files sorted in various time, energy and detector domains. After initial processing, the files associated with a single burst or solar flare become known collectively as an Individual Burst Data Base (IBDB). These IBDB's, which are about 1 Mbyte in size, are made available to the BATSE Co-Investigators and selected Guest Investigators. The IBDB's are available from the public archive a year after production. A Fortran program BFITS is available for extracting from an IBDB, and writing to a FITS file, a time series of count spectra constructed from any of the available data types. An IDL based package called WINGSPAN has been developed to perform spectral analysis of burst and flare data contained in the BFITS output files. Additional burst analysis software exists to determine burst locations, perform temporal analyses, display the temporal evolution of spectral parameters, and derive parameters for global statistical studies.

    Pulsar analysis is a secondary BATSE science objective. Daily FFT's of DISCLA data are performed to detect transient pulsed sources. The collected folded-on-board data is phase aligned and refolded to produce statistically significant pulse light curves. In addition CONT and DISCLA data is folded on the ground for longer period sources. Time series of pulsed light curves are correlated with pulse shape templates to determine pulse times of arrival. These are used in studies of binary orbits and pulsar rotational dynamics. Software has been created for light curve display, time-resolved spectroscopy, and display of temporal evolution of fit parameters. A pulsar source catalog is maintained and will be made available for public access.

    The monitoring of persistent and long-term transient sources using Earth occultation is also a secondary BATSE objective. Pipeline processing of the CONT data is performed for quick-look science analysis. Analyses using more refined models are then conducted. Software exits to produce long term light curves and flux spectra from the CONT and to some degree from the DISCLA and DISCSP data. High level data products also exist for more complicated analyses. The software has evolved considerably since the start of the mission. To date 35 sources have been detected.

    Studies are being conducted of the BATSE detector background. These studies have led to modeling techniques for subtracting the background from the CONT and DISCLA data. This model-subtracted data is expected to be useful in pulsar studies, for searches for non-triggered bursts or intermediate-timescale transient events, and for improving the sensitivity of source observations by Earth occultation, and is currently being tested for these purposes.

    BATSE can attain some of its scientific objectives without long observation times or extensive data analysis. For example, the location of bursts and their spectra and detailed time history are available soon after their occurrence. Similarly, new and recurrent transient sources can be rapidly detected so that prompt follow-up observations with other instruments may be possible. In order to implement the "quick-science" capabilities of BATSE, the entire BATSE data stream is examined soon after receipt. BATSE maintains a "Quick-Alert" distribution network for providing rapid notification and coarse locations of gamma-ray bursts to collaborators and approved guest investigators. A BATSE network for near-real-time burst notification, BACODINE, has been implemented by collaborators at GSFC (Barthelmy,1994). A solar flare monitor based on BATSE data and a BATSE solar flare data archive are maintained by the Solar Data Analysis Center at Goddard Space Flight Center. This service can be reached via Internet by logging into sdac.gsfc.nasa.gov with username "BATSE."

    D. Data Analysis Environment

    The BATSE data analysis environment at MSFC consists of workstations and mass-storage devices including a read/write optical disk jukebox. Workspace for several short-term visitors is available.

    E. BATSE Guest Investigator Opportunities

    Opportunities for guest investigations exist in the areas of -ray burst studies, discrete source monitoring, pulsar observations, and solar observations.

    Burst Studies.

    All burst data over a year old is available through the archive at the GRO-SSC. Data from most recent gamma-ray bursts are made available to guest investigators by the BATSE team. The most recently released catalog of bursts detected by BATSE is available through the GRO-SSC. A CD-ROM containing data from all bursts in this catalog, together with useful software is expected to be available in early 1995. Investigations that depend upon the entire catalog of -ray bursts, prior to publication, may be inappropriate.

    Studies to attempt to identify burst counterparts are encouraged. Burst locations are routinely determined within 24-48 hours of occurrence. Coarse locations are available in 4 seconds through the Global Coordinates Network (GCN) <http://lheawww.gsfc.nasa.gov/docs/gamcosray/legr/bacodine/gcn_main.html>. Also refer to Barthelmy et al (1994).

    Comparison of BATSE locations with interplanetary network solutions, hard solar flares, and Cyg X-1 fluctuations indicate that the average systematic location error is 4 (Fishman et al. 1994).

    Discrete Longer-Lived Sources and Pulsars

    The Compton-BATSE experiment has considerable capabilities for observations of steady sources, pulsed sources and longer-lived transient sources. (Fishman, et al. 1989, Harmon et al. 1994, Prince et al. 1994). A list of known sources that may be observable by BATSE occultation or periodic emission analysis is available. From this list, approximately 60% of the sources will be available for guest investigations. Observations of short period pulsars would need to be scheduled into the on-board epoch-folding data system (Wilson et al., Isolated Pulsar Workshop, Taos, NM, 1992). It is possible to observe pulsars with periods as short as a few milliseconds. Longer-period pulsars (>10 sec) need no special mission planning. Data can be analyzed either at a BATSE institution or the guest investigator's institution.

    Solar Observations

    Numerous opportunities exist for detailed studies of individual flares from solar flare-triggered BATSE data. For those flares that trigger the BATSE burst data system, hundreds of spectra are accumulated during the flare onset. During this time interval, time resolutions of tens of microseconds should be available with some limited spectral information. This should put severe constraints on the spatial distribution and evolution of high energy electron populations within the flare region. In addition to the electron component of flares, the high energy ion component from several MeV to several GeV can be studied through the emission of -ray lines, pion decay, the decay of positron-emitting isotopes, and neutrons. Each of these species will have different temporal relationships with the flares, permitting additional insight into the spatial and temporal characteristics of the high energy ions within the flare regions.

    F. BATSE Flight Performance and Observation Sensitivity

    Gamma-Ray Bursts. The in-flight performance of BATSE for detecting -ray bursts is at least as good as the pre-launch predictions. A trigger sensitivity of about 10-7 erg/(cm2s) is typically achieved for most bursts. The burst occurrence rate at the normal threshold settings is slightly less than one per day. The rate of on-board triggers is significantly larger than this, averaging about 2.5 per day. At the beginning of the mission many of these were solar flares (see below). Nearly all the others are attributed to the effects of electron precipitation. Events may occur locally, producing -ray secondaries in the Compton spacecraft which trigger BATSE, or as auroral events at some distance from the spacecraft. The two types of events leave distinctive signatures in BATSE and are not easily confused with -ray bursts.


    Hard X-Ray Source Monitoring. Using the full unocculted sky viewing of the BATSE detectors, a significant contribution can be made to monitoring a large number of hard x-ray sources, many of which are variable, much like the contributions made by GINGA and Ariel-V at somewhat lower energies. While the BATSE detectors do not have the sensitivity and energy resolution of the collimated, pointed OSSE experiment, they do have the capability to provide frequent samples of the intensity of moderately strong steady sources, using the Earth occultation technique. This is demonstrated in Figure V-2 which shows occultation edges in the DISCLA data due to Cyg X-1.

    The BATSE team's current approach to monitoring bright sources is to obtain a best fit on several minutes of background before and after a predicted occultation time. The current fitting model uses an atmospheric transmission function for the source contribution combined with a background which is quadratic over the sampling interval. Using the detector response matrices appropriate to the detected source direction, the flux corresponding to the estimated counts is obtained.

    Occultation steps in DISCLA channels 1 and 2 due
to Cyg X-1

    Figure V-2. Occultation steps in DISCLA channels 1 and 2 due to

    Cyg X-1. The data are for detector 6 on May 8 1991.

    To monitor weaker sources the sensitivity of the method is enhanced by combining data from multiple orbits. For daily mission operations and known source locations, the current approach is to fit individual occultation edges and combine these measurements, usually over a period of one day. Intervals of poorly-behaved background due to incidental events such as flares or bursts are excluded. The step fitting process is then applied to each of the 16 energy channels of the CONT data. The photon flux is obtained by fitting the steps to a power law, and both the flux and the source count rates are stored for each source. The 56 sources currently being monitored are listed in Table V-3.

    Table V-3.

    Sources Currently Monitored by the Occultation Method

    Crab SMC X-1 Her X-1 Cen X-3 4U 0115+634 4U 1627-673
    1E 1024-57 Cyg X-1 NGC 1275 A 0525+262 GX 301-2 PSR1509-58
    NGC 4151 3C 273 Cir X-1 Sco X-1 GX 339-4 GRO J1008-57
    4U 1700-337 GX 1+4 Aql X-1 Cyg X-3 Vela X-1 Cen A
    3C 279 1E 1740-29 A 1118-616 OAO1657-415 EXO2030+375 GX354+0
    A0620-00 Nova Mus1991 GRS 0834-430 GX 5-1 Nova Sag 1993 WR 140
    GRS 1758-258 Geminga MCG 8-11-11 Mrk 421 Nova Cyg 1992 Cyg X-2
    GS 2000+25 4U 1543-45 EXO 1846-031 GRO J0422+32 GRS 1915+105 NGC 5548
    PKS 2155-30 SN 1993J SCT X-1 GRO J1009-45 GRO J1719-24 SN 1994W
    GRO J1655-40 OJ 287 * * * *

    BATSE steady source detection sensitivity, using
the occultation method

    Figure V-3. BATSE steady source detection sensitivity, using the occultation method. A Crab-like energy spectrum is assumed.

    Searches for unknown sources are routinely performed by a program which detects step features in the data using a fit over a sliding time interval. Locations of candidate sources are determined from the intersection of the projection on the sky of the Earth's limb at source rise and set. Sky imaging of the Galactic plane and surrounding regions on a routine basis is currently being explored.

    BATSE's sensitivity using the occultation technique with the LADS is shown in Figure V-3. Currently only LAD data are being analyzed on a daily basis. The Crab spectrum is shown for reference. The Figure shows the sensitivity of BATSE for source detection based on the use of individual CONT data channels. Improved sensitivity can be achieved by using multiple channels. The sensitivities are calculated from a two-week integration of observed Crab occultation steps and an assumed Crab spectrum (Jung, 1989).

    Pulsed Source Observations. BATSE routinely makes observations of pulsed sources, using both the on-board folding hardware (FOB), and by FFT-based or epoch-folding analyses of the 1.024 s resolution, 4 energy channel Discriminator data and 2.048 s resolution, 16 energy channel Continuous data. The objects for which FOB data has been collected as of 16 September 1992 are shown in Table V-4. Those sources marked by an "*" have been detected in the data; those marked by a "#" are "single-sweep" (i.e. non-folded) data, typically 31 ms resolution.

    Table V-4

    Sources For Which On-Board Pulsar Data Has Been Acquired

    Crab Pulsar* Vela Pulsar PSR 1509-58* Her X-1*
    Cen X-3* SMC X-1 Geminga GX 339-04#
    GRO J0422+32# GS 0834-43# 4U 1627-673 4U 0115+634
    A0535+262# SGR 1806-20# GC(-40,0)# GC(-30,0)#
    GC(-20.0)# GC(-10,0)# GC(0,0)# GC(10,0)#
    GC(20,0)# GC(30,0)# GC40,0)# PSR 0355+54
    PSR 0329+54 PSR J0437-471 PSR 0525+21 PSR 0540-693
    PSR 0611+22 PSR 0628-28 PSR 0656+14 PSR 0736-40
    PSR 0740-28 PSR 0818-13 PSR 0823+26 PSR 0835-41
    PSR 0950+08 PSR 0959-54 PSR 1055-52 PSR 1133+16
    PSR 1237+25 PSR 1259-63 PSR 1508+55 PSR 1556-44
    PSR 1641-45 PSR 1642-03 GRO J1655-40# PSR 1706-44
    GRO J 1719-24# PSR 1737-30 PSR 1749-28 PSR 1800-21
    PSR 1813-36 PSR 1818-04 PSR 1822-09 PSR 1823-13
    PSR 1831-03 PSR 1839+56 PSR 1919+21 PSR 1929+10
    GRO J 1948+32# PSR 1951+32 PSR J 2043+27 PSR 2334+61

    Figure V-4. Crab pulse profile obtained with 1.5 x 105 s exposure during TJD 9615-9629.

    the light curve obtained for the Crab Pulsar

    Figure V-4 shows the light curve obtained for the Crab Pulsar for a 1.5 x 105 s exposure. The data was collected over a period of 14 days, with about 1/4 of each GRO orbit devoted to this source.

    Figure V-5. Sensitivity of BATSE for detection of pulsed sources, given as the Root Mean Square (RMS) deviation of the source flux from its phase-averaged mean.

    BATSE's sensitivity for pulsed source detection

    Figure V-5 shows BATSE's sensitivity for pulsed source detection. The sensitivity is presented in terms of the Root Mean Square (RMS) deviation of the source flux from its phase averaged mean. For reference the RMS pulsed flux of the Crab pulsar is shown. The 3 sigma sensitivity for detection using individual CONT channels are shown (points marked by diamonds). The better sensitivity that can be obtained by combining channels is also shown. The sensitivity assumes a 100,000 s livetime, which for folded-on-board data takes 8-16 days to acquire, depending on the fraction of the orbit devoted to the source. The sensitivities were calculated from a 12 day observation (156,000 s livetime), using 2 BATSE LADs, exposed at angles of 44 and 42 degrees, respectively, a representative exposure geometry. For pulsars of unknown or poorly known period, the calculation of detection confidence must incorporate the number of independent frequencies searched, in addition to the flux sensitivities shown here.

    The DISCLA (1.024 second resolution) and CONT (2.048 second resolution) data types are obtained whenever the high voltage is on, so that searches for pulsed emission from long-period x-ray binary pulsars are possible whenever the source of interest is not occulted by the Earth. To date the pulsars detected by FFT or epoch folding analyses of these data types are Her X-1, 4U 1626-67, GX 1+4, Cen X-3, OAO 1657-415, Vela X-1, 4U 1538-52, GX 301-2, 4U 0115+62, 2S 1417-624, EXO 2030+375, A 0535+26, A 1118-616, 4U 1145-619, GS 0834-430, GRO J1948+32, and GRO J1008-57.
    Solar Observations. Solar activity was fairly high early in the Compton mission. The average rate of BATSE triggers due to solar flares has dropped from about one per day in 1991 to about one per 10 days in 1994. Since no attempt is made to bias BATSE against these triggers, the sensitivity for solar flare triggers in the 50-300 keV range is the same as for cosmic bursts. Unlike cosmic bursts, flares have relatively soft spectra; hence, many non-triggered flares are visible in the LAD data below 50 keV. The lowest discriminator channel in the SDs has also proven to be a valuable monitor for solar activity at energies as low as 5 keV. At least one SD with a front hemisphere view of the Sun has been kept in the high gain mode for this purpose. GOES data are routinely used to confirm the origin of suspected solar flare triggers. Many of the BATSE triggers are associated with events of GOES x-ray class C2 or weaker.

    A number of intense flares (GOES x-ray class X1 or higher) have occurred since the Compton launch and have been detected by BATSE. However, the high sensitivity of the BATSE detectors, particularly at low energies, combined with the optimization of the trigger memory accumulation for bursts, results in some significant instrumental limitations for very strong flares. Such flares typically produce rates in the range of 200,000 counts/s in each of the Sun-facing BATSE detectors, requiring significant corrections for electronics dead time and pulse pile-up. Although an effort to understand all of the dead time effects in BATSE is in progress, no study of pulse pile-up effects has yet been undertaken. The persistence of such high count rates for several minutes or more is also a problem for the burst memory storage. Most of the burst data types are filled after only a short interval. The use of non-burst data types generally requires careful corrections for counter overflow as well as the above-mentioned problems. Even detectors not directly facing the Sun are adversely affected during the most intense flares as a result of atmospheric scattering.

    G. Defining an Observation

    Unlike the other instruments on the Compton spacecraft, BATSE has little inherent angular resolution. Data may be selected which is optimal for a specific source, but the flux from any steady source is invariably confused with that from many others. BATSE can only be made sensitive to a single source via timing information, as with burst sources, solar flares, pulsars, or the Earth's occultation of steady sources. These observations are defined by selection of a data type, times of observation and possibly a pulse period. Thus the observer's definition of an observation for BATSE will not affect the Compton spacecraft orientation. The definition may in some cases affect the BATSE operational configuration, and will in all cases affect the selection of the subset of data to be given to the guest investigator for analysis.

    Guest investigators who are requesting BATSE data should fill out the Observation Definition Form, included in Appendix D, to define the requested observations. The purpose of this form is to obtain from the proposer a definition of the data set needed for the investigation. If the form is inadequate for this purpose, the proposer should not feel constrained by its structure - text can be appended in the box labeled "special requirements."

    In filling out this Form, the following guidelines apply:

    1. At least one separate row should be used for each target or class of targets requested.

    2. For mode specify "Triggered" for triggered burst and solar flare data, "Pulsar-FOB" for hardware folded pulsar observations, "Background" for DISCLA, CONT, and normally scheduled SHER and HER data and "Occ-hist" for occultation histories. Note that some of these fields may get truncated on the electronically generated forms, but they will be properly translated into the observation data base.

    H. Data Products.

    Low level Data.

    Individual Burst Data Base (IBDB). This includes time of trigger and derived source location, and all associated burst mode data (DISCSC, DISCLB, PREB, MER, TTS, TTE, STTE, HERB, SHERB as collected) filtered for data quality, for a given burst or solar flare. Relevant background data (DISCLA, CONT, HER, SHER), associated spacecraft position and attitude information, and energy calibration information is included. It is recommended that investigators used the BFITS software to produce extract time histories of count spectra from the IBDB's rather than reading them directly.

    DISCLA or CONT FITS files. These contain DISCLA or CONT data with associated spacecraft position and calibration information.

    HER_COR or SHER_COR files. These contain overflow corrected HER or SHER data.

    Pulsar Low Level FITS files. These contain unpacked and overflow corrected on-board integrations, at the original time resolution. They include computed Solar System Barycenter (SSB) arrival time at midpoint of each on-board accumulation and associated spacecraft position and attitude, plus calibration information.

    High-Level Processed Data Products.

    Triggered Count Spectra Histories. These products, which are produced from the IBDBs by the program BFITS, give the time history of count rates in a set of energy range for a single burst. They are derived from DISCSC, PREB, TTS, DISCLA, MER, CONT, DISCSP, STTE, HERB, HER, SHERB, SHER, or STTE data as appropriate. It includes the trigger time and derived source location and calibration information.

    Source Flux Histories. This product gives the time histories of the integrated flux of monitored sources by the occultation technique. For 16 channel data it includes the derived change in counting rates at occultation. Calibration information is also included.

    Occultation Fit Data Set. This product contains data associated with the occultation steps caused by a source or group of sources. It includes times of occultation edges, count rate histories near edges (medium energy resolution), model fits of the steps, associated spacecraft position and attitude, and calibration information.

    Pulsar Epoch Folded Data Set. This product consists of folded-on-board or modeled CONT data epoch folded to obtain statistically significant pulsed light curves for a single source. It includes phase and timing model information and calibration information.

    Catalogs.

    Burst Catalog. This catalog summarizes information about the bursts BATSE has detected giving the standard name, BATSE ID number, trigger time, derived location and derived parameter values for each burst.

    Pulsar Observation Log. This contains information on the data that has been collected using the on-board epoch folding hardware.

    Software.

    BFITS. Software to read IBDBs and produce FITS formatted Triggered Count Spectra Histories.

    DRM_GEN. Software to generate response matrices for the LAD and SPEC detectors in a FITS format.

    WINGSPAN. IDL software for burst temporal and spectral display and spectral analysis. This software uses as input the FITS files generated by BFITS and

    DRM_GEN.

    Documentation for these software products are available.

    References

    Bunner, A.N., "The GRO Guest Investigator Program", GRO Science Workshop, GSFC, 1989.

    Barthelmy, S.D., Cline, T.L, Gehrels, N., Kuyper, J.R., Fishman, G.J, Kouveliotou, C., and Meegan, C.A., "BACODINE", in Gamma-Ray Bursts, AIP Conf. Proc. 307, 643-647, 1994.

    Fishman, G.J., Meegan, C.A., Wilson, R.B., Paciesas, W.S., Parnell, T.A., Matteson, J.L., Teegarden, B.J., Cline, T.L., Pendleton, G.N., and Schaefer, B.E., "The Burst and Transient Source Experiment (BATSE) - Scientific Objectives and Capabilities", GRO Science Workshop, GSFC, 1989.

    Fishman, G.J. et al., "The first BATSE Gamma-Ray Burst Catalog", ApJ Suppl. 92,229-283, 1994.

    Harmon, B.A., Zhang, S.N., Wilson, C.A., Rubin, B.C., and Fishman, G.J., "BATSE Observations of Hard X-ray Sources", Proc. Second Compton Symposium, AIP Conf. Proc. 304, 210-219, 1994.

    Horack, J.M., Development of the Burst and Transient Source Experiment (BATSE), NASA Reference Publication 1268, September 1991.

    Jung, G.V., "The Hard X-Ray to Low-Energy Gamma-Ray Spectrum of the Crab Nebula", Ap.J. 338, 972-982, 1989.

    Mahoney, W.A., Ling, J.C., and Jacobson, A.S., "HEAO 3 Observations of the Crab Pulsar", Ap. J. 278, 784-790, 1984.

    Prince, T.A., Bildsten, L., Chakrabarty, D., Wilson, R.B., and Finger, M.H., "Observations of Accreting Pulsars", in The Evolution of X-ray Binaries, AIP Conf. Proc. 308, 235-244, 1994

    VI. THE COMPTON GRO SCIENCE SUPPORT CENTER

    A. General

    The Compton Gamma Ray Observatory Science Support Center (SSC) was established to provide the central contact point for Guest Investigators at all stages in their use of Compton resources. The SSC provides the main interface between the Guest Investigator community and the other aspects of the CGRO project. In particular, the SSC provides assistance with preparing proposals, retrieving and analyzing data, maintains the archive, and integrates data products into the other NASA facilities (such as the HEASARC - High Energy Astrophysics Science Archive Research Center).

    The SSC comprises scientists and support personnel who are specialists in analyzing data from the Compton instruments and provides hardware and software capabilities to facilitate research activities. The SSC is located at Goddard Space Flight Center in Greenbelt, Maryland. Address any inquiries regarding the Guest Investigator program, products or analysis software to CGRO-SSC.

    B. Guest Investigator Support

    The SSC will provide support to Guest Investigators at all stages of Compton research from the proposal generation through archival research. A good starting point for many inquiries is the CGRO WWW site: <http://cossc.gsfc.nasa.gov/>. Alternatively, e-mail any enquiries to CGRO-SSC.

    The SSC is located at the Goddard Space Flight Center. The SSC provides direct support for astronomers, maintains the data archive, maintains a CGRO WWW site and other on-line services, and provides a single point of contact for any mission-related questions. It is a useful starting point for many CGRO inquiries. The SSC has the capacity to support several Guest Investigators on-site and is particularly well suited to archival and multi-instrument research. Staff at the central facility are knowledgeable about all of the Compton instruments and can refer to the Instrument Specialists or Instrument Teams when detailed questions arise. In addition, the facilities of the HEASARC, NSSDC, and other scientific support organizations located at Goddard can be useful in support of multi-mission and multi-disciplinary research.

    Facilities:

    Computer facilities for the use of Guest Investigators are available at the CGRO-SSC office and each of the Instrument sites. The instrument team sites are described in the individual instrument sections of this Appendix. The SSC maintains a diverse array of hardware platforms and operating systems, which allows support of the various instrument specific software systems. The SSC can provide users interested in CGRO data analysis with accounts on the SSC computer systems.

    SSC Analysis Software:

    In order to keep pace with archival users, and to serve those with current programs, the SSC engages in a continual effort to port instrument team software and develop local tools. Software to convert data products from their native format to FITS format and back again, software to control and access the data archive, the development of a standard Graphical User Interface for all GRO software, and efforts to create quick-look tools are examples of software development relevant to the central facility beyond the porting of IT software. Refer to section C below for information on software utilities useful for accessing the data archives.

    In addition to GRO specific software, other common packages are available for use at the SSC. IDL, XSPEC, and the HEASARC "F-tools" library are examples of software which can or are currently being used to make use of GRO data products. For example, quick-look tools being developed at the SSC using IDL are a valuable resource for rapidly assessing data for a particular program. In general, instrument team software is available at the central facility or at the IT sites. Archival researchers are encouraged to use the SSC facilities. If use at a IT site is required, accounts will be set up on computers by each PI team to allow Guest Investigators to access the facilities remotely. These accounts will be managed jointly by the SSC and the IT and will provide users with access to the standard instrument data analysis software. On-line documentation will be available at each site. Users should always feel free to contact either the SSC central office or the IS when they have questions not answered in the documentation.

    Most Instrument Team software packages are available at the CGRO-SSC. For instance, the IGORE (Integrated Gamma-Ray Observatory Reduction Environment) package from the OSSE team provides everything needed for the analysis of OSSE data. This package, which relies heavily on IDL, allows the user to sum spectra, perform background subtraction, fit models, and perform pulsar analysis. The COMPASS (COMptel Processing and Analysis Software System) is installed at the SSC as well. The BSAS (BATSE Spectral Analysis Software) and WINGSPAN (WINdows Gamma SPectral ANalysis) packages are available at the Science Support Center for spectral analysis of BATSE burst data. Several of these packages, IGORE and BSAS, are available for installation at a users own institution, however they are integrated with other software (i.e., INGRES, IDL, ORACLE) which require additional commercial licenses. A simpler solution in many cases may be to request an account on the SSC's computer systems.

    Software is now available for the EGRET instrument to accomplish everything from pulsar analysis (PULSAR) to creating your own skymaps based upon individual event selection (INTMAP, MAPGEN) to viewing maps and event lists (QUICKLOOK, SHOW). Other EGRET software such as LIKE, which performs a maximum likelihood analysis of possible point sources, is available as well. This software requires a model of the diffuse galactic background radiation. The model currently installed at the SSC is version 5.3 of Bertsch et al. (1993, Ap.J. 416, 587).

    Contact the CGRO-SSC to find out what software is currently available -- or browse our WWW site at <http://cossc.gsfc.nasa.gov/>.

    C. Data Catalog and Archives.

    Non-proprietary data from the Compton Observatory is available through the Compton Observatory Archive. This archive is managed by the SSC and is open to astronomers for use in archival research, proposal preparation or for any other scientific purposes. Data is placed in the archive when delivered to the SSC by the instrument teams, nominally one year after the data is available to the team (or the Guest Investigator for data from GI observations). Any user may ask for non-proprietary Compton data, but if significant resources will be needed an astronomer should submit an archival research proposal. The archive can also be used to supplement observations obtained in the course of an observational proposal.

    1. Data Contents.

    The data types which are to be stored in the archive are detailed in the CGRO PDMP(1) (Appendix H). Both low-level and high-level data types are stored in the archive.

    2. Archive Data Catalog and User Interface.

    The SSC has developed a catalog of the contents of the CGRO archive and provides network access to both the archive and catalog. The catalog indexes the archive by position, instrument, data type, time of observation, target and target type. Users may select data from the archive according to one or more criteria. Once a user has selected the data of interest he or she may ask for this data to be retrieved from the archive. Normally a user will then transmit the data to his or her home institution over electronic links. However, if the volume of data is too large, or the user's network links inadequate, the SSC will generate and mail physical media.

    A graphical user interface to the archive selector is accessible under the HEASARC W3-Browse utility, accesible via the CGRO-SSC home page on the World Wide Web (WWW) URL: <http://cossc.gsfc.nasa.gov/>.

    Data and documentation on is available through anonymous FTP in /pub/data or /pub/doc/archive on `cossc.gsfc.nasa.gov' or on the WWW under the CGRO home page. Data can also be retrieved through the HEASARC on legacy.gsfc.nasa.gov under /compton/data.

    3. Archive Data Formats.

    All data stored in the CGRO archive is stored in FITS formats. Most data uses the binary tables extension to FITS. Some data types are available at the SSC but not yet converted to FITS. These data may be obtained by contacting the SSC. Further documentation of the data formats used in CGRO data may be obtained by contacting the SSC. Users do have the option of retrieving data in a compressed format (as generated by the UNIX compress command). This may be helpful in retrieving large amounts of data. Typically CGRO data files compress by significant factors.

    4. Archive Data Distribution.

    Currently, interfaces to the CGRO archive retrieve the data to stagin disks at the CGRO-SSC. It is anticipated that users will then retrieve the data over the Internet using FTP or DECnet copy protocols to get the files to their home institutions. Note that all FITS files should be retrieved as binary files. In response to user requests the CGRO-SSC staff can mail physical media to investigators. Currently, 8 and 4 mm cassette tapes are the preferred media for data distribution.

    5. Support for Archival Research

    The SSC will act as the primary site to support archival research by astronomers using Compton data. Visitors can be provided with office space if they wish to work at our facility to a limit of three visitors at any one time. The SSC will also support archival researchers who wish to work at their home institutions. The SSC will set up computer accounts and provide disk space on the SSC cluster for either type of researcher. Users who wish to visit the SSC should contact the SSC staff to arrange an appropriate time for the visit. Any archival researcher who has questions should feel free to contact the SSC to discuss their problems.

    In addition to the SSC developed or supplied software facilities, the full capabilities of the HEASARC are available to CGRO archival investigators. This includes data from many other high-energy missions including ROSAT, Einstein, EXOSAT, ASCA, SAS-2, and COS-B.

    6. EGRET and BATSE CD ROMs