HEAO-1 and the A2 Experiment
The Satellite, Mission, Experiment, and Data
The first High Energy Astrophysical Observatory (HEAO-1) was primarily a survey mission, dedicated to systematically mapping the X-ray sky every 6 months. The satellite was launched on 12 August 1977 into a nearly circular orbit with apogee 445 km and inclination 22.75°. HEAO-1 had a 93 minute orbital period, and, while in scanning mode, spun with a nominal period of 33 minutes. Each spin traced out a great circle of constant ecliptic longitude. Every twelve hours, the spin axis was moved approximately 0.5 degrees in order to keep it pointed at the Sun; thus, after 6 months, the entire sky had been observed. After the first ~ 100 days of the mission, scanning was interrupted from time to time to point the detectors at particular objects of interest. These pointing times became more frequent until 9 January 1979, when the gas used to control the spacecraft attitude ran out. The systems were shutdown, and HEAO-1 drifted in a decaying orbit until March 1979, when it burned up on re-entry into the atmosphere.
The A2 experiment (a collaborative effort led by E. Boldt (GSFC) and G. Garmire (Cal Tech/PSU)) was designed to study the large scale structure of the galaxy and the universe, yielding high quality spatial and spectral data in the X-ray region. However, the detectors also produced valuable information on discrete X-ray sources such as binary star systems, hot white dwarfs, cataclysmic variables, and supernova remnants (see, for example, Cordova et al. 1984; Nugent et al. 1983). Extragalactic objects such as radio galaxies, Seyfert galaxies, cluster of galaxies, quasars, and BL Lac objects were also well-studied (see, for example, Piccinotti et al. 1982; Worrall et al 1981; Mushotzky 1984). For its time, the A2 experiment produced the best spectra ever obtained over the 2-60 keV energy range.
The experiment consisted of 6 separate multi-anode, multi-layer, collimated gas proportional counters covering 3 different energy ranges (Rothschild et al. 1979). Two of the detectors, designated LEDs (Low Energy Detectors), were thin window propane-filled proportional counters sensitive to X-rays from 0.15 - 3.0 keV, and having about 400 cm2 open area each. There was one MED (Medium Energy Detector) which consisted of an argon filled counter covering the energy range 1.5-20 keV. Finally, there were 3 HEDs (High Energy Detectors), which were xenon filled counters covering the range 2.5 - 60 keV. The MED and the 3 HEDs had roughly 800 cm2 open area each.
The HEDs and MED had various combinations of 1.5° x 3°, 3° x 3° and 3° x 6° fields-of-view (FWHM). The collimators were oriented so that the 3° angular response was always perpendicular to the scan plane. Thus, each rotation of the satellite scanned a great circle 3° wide on the sky passing through the ecliptic poles.
The HEAO-1 time is defined in days of 1977. Launch occurred on 12 August 1977 or day 224. The first week of operation was devoted to turning on the detectors and calibrating their performance. At day 232.04, the detectors' high voltages were changed to their normal states where they remained for the remainder of the mission. Overall detector stability was achieved for the HEDs on day 234.95; MED stability was achieved on 248.15.
In-flight calibration for each detector used an internal Fe55 source which illuminated the end veto cells. The count rates were low, so the rejection of charged particles would not be compromised. The MED also had an Fe55 source mounted on a rotating assembly that could be moved into the field-of-view (FOV). The HEDs also contained an Am241 source, located between two anodes behind the last layer of the detector. The Am241 source was weak enough that interference with science data was negligible. Finally, electrons incident on the copper in the HED collimators produced a characteristic 8.14 keV line which was also used for calibration. Calibration data were recorded for all detectors continuously, though there were also times devoted to calibration.
All of the satellite data, regardless of format or quality, were placed in the MAX database. It was reformatted from telemetry stream order to observation time order, with all the data from a single major frame (40.96 s) accumulated in a single tape record. The standard data formats were unpacked and preliminary flags set for general data quality, engineering/science format, earth occultation, electron contamination, magnetic field direction, high voltage, and SAA/NPA passage. Data were accumulated in spectral windows and combined by the spacecraft into broad energy ranges. These summed rates were referred to as 'discovery scalars' and were read out every 1.28 s. Thus, a major frame consisted of 32 consecutive sets of discovery scalars. More will be said about these in a later section.
Several subset databases were generated from the MAX data to speed access and reduction of the data. These include the PHA, XRATES, and DSDISK databases:
* The PHA database contains a complete set of the Pulse Height Analyzed (PHA) and some discovery scalar data, as well as the various data flags described above. Only major frames in which at least one detector had useful data are included.
* The XRATES database is similar to the PHA database, except that it contains a complete set of rates rather than PHA data. Again, the various data flags are included in this database.
* The DSDISK database is a subset of XRATES. There are 2 versions: one for the 3° x 3° and one for the 3° x 1.5° FOV data. The data spans days 232.1 - 737.1 (the full lifetime of the mission) for the MED and HED3 detectors.
What is Online Now?
Currently online at the HEASARC, you can find the following A2-related files:
* A catalog of sources observed by A2 with all 3 detectors during its pointed observations can be found in BROWSE as A2POINT or can be found in README.heao_pointings in /FTP/heao1/data on legacy.
* The Piccinotti catalog can be found in a BROWSE database, A2PIC. It contains data for 66 non-galactic sources either brighter than 1.25 R15 counts/sec in the first scan (1977 day 248 to 1978 day 72), or brighter than 1.8 in the 2nd scan (1978 days 73 to 254). The identified sources fall into several categories, including: narrow emission line galaxies, broad emission line galaxies, BL Lacertae objects, and clusters of galaxies.
* The HEAO-1 A-2 LED Catalog of High-Energy X-ray Sources can be found in BROWSE as A2LED. This catalog is the result of a study of the diffuse X-ray sky over the bands of X-ray energies 0.18-0.44 keV and 0.44-2.8 keV from August 1977 until January 1979 using data obtained with the A-2 Low Energy Detector. Using a significance criterion of 6 sigma for existence, 114 sources are cataloged.
* A bibliography of publications from, and about, HEAO-1
* SkyView, a forms-based WWW program to make selected maps from a broad base of source catalogs, currently holds the HEAO-1 A2 all-sky data in the Total X-ray color (2.0 - 60.0 keV) with the SFOV (3.0° x 1.5°) collimator. The SkyView data spans mission days 322 - 502 and has been processed using the SAMPLED alogithm described below. The WWW address for SkyView is http://skview.gsfc.nasa.gov/. SkyView can automatically transform the data into any coordinate system and epoch, as well as displaying it in any of a large number of possible projections.
* All-sky maps from the LED during the survey phase, and spectra from objects observed during the pointing phases which were delivered to us by Dr. John Nousek (Pennsylvania State University) as part of a NASA Astrophysics Data Program grant.
* All-sky maps from the MED and HED3 detectors during the survey phase. The maps are available for several coordinate systems, projections, and X-ray colors, both for the large (3.0° x 3.0°) and small (3.0° x 1.5°) FOV. These maps and formats are described in considerably more detail below.
* Lightcurves and spectra for over 400 sources observed by A2. As of this time, these files are yet to be made, but their delivery is expected before, or soon after, the publication date of this Legacy issue. These files will be in standard OGIP FITS format.
About the MED/HED3 A2 All-sky Maps
We have recently added the all-sky scanning data from the MED and HED3 detectors of the A2 experiment to the HEASARC online database. Maps are available in 4 separate colors, in both large (3.0° x 3.0°) and small (3.0° x 1.5°) FOVs, and covering 2.5 complete scans of the sky.
Our philosophy has been to provide as much data as practical, in as raw a form as practical, while simultaneously providing maps which have been cast into useful (i.e. familiar) coordinate systems. These latter products will allow members of the astrophysical community to quickly examine the data to "see what is there" and to access the data in a format where the projection and statistical uncertainties are easily and completely described. This second capability is desirable due to the large FOV.
To use the maps, the user should be aware of how data were combined from the spacecraft and how certain keywords in the FITS files are defined. Otherwise, images like Figure 1, which covers a time period during which detector changes occurred, may seem somewhat disconcerting! In particular, the user should be aware of the definitions of the discovery scalars and X-ray colors.
The Discovery Scalars
The discovery scalars are a set of combinations of spectral windows in the HEAO-1 A2 detectors. The scalars were defined by commands to the spacecraft, and those combinations were altered several times during the HEAO-1 mission (though most alterations were during the first weeks of the mission). For the MED and HED detectors, there were four spectral windows in the first layer of the detectors and two in the second layer. Table 1 (Marshall 1983) presents the nominal prelaunch values. Note that the windows are defined as voltage levels in the analog processing; small variations in the detector gas gain will shift the edges of the windows (in keV).
Each detector produced four pairs of rates, or discovery scalars, every 1.28 s. The pair members differed only by their FOV. The spacecraft combined data from the spectral windows to make these discovery scalars using combinations listed in Table 2. The definitions were changed 5 times from the initial state: the last column in Table 2 provides the times each mode was in use.
Mode HED Windows MED Windows Days of 1977
Most users of the HEAO-1 A2 data will not want to use the discovery scalars directly. Instead, they will want to work with certain well-defined standard combinations of the scalars referred to as X-ray colors. Four colors are commonly used: Hard, Soft, Total, and R15.
Because the discovery scalars were changed during the mission, there are sharp discontinuities in the total, hard, and soft bands. The R15 color is a sum of discovery scalars which did not change during the mission. In principle, changing the weights for summing the scalars would take out the apparent abrupt changes in sensitivity. In practice, however, the discovery scalars cover broad energy bands (see below) and the correction can only be done for a single input spectrum. The weights have been adjusted to minimize variations in measured rates for sources with a "canonical AGN spectrum" (i.e. a photon index of -1.7 (Mushotzky 1984)). Table 3 shows the standard weights used to define the four X-ray colors.
Note that for R15, the rate is just the unweighted sum of the first and second layers of the HED3 and the second layer of the MED (Marshall et al. 1979). This band has been widely used in A2 analyses of fluxes (Piccinotti et al. 1982; Della Ceca et al. 1990); diffuse galactic emission (Worrall et al. 1982); and auto correlations studies (Persic et al. 1989; DeZotti et al. 1990). Recent work has shifted to the total band which has higher statistical precision and a lower mean energy due to inclusion of the first layer of the MED (Jahoda and Mushotzky 1989; Lahav et al. 1993; Miyaji et al. 1994; Boughn and Jahoda 1993).
Since the colors are combinations of several spectral windows, the MED and HED effective cross-sectional areas for the colors are somewhat complicated. Figures 2a, 2b, 2c, and 2d show the approximate effective area for the four colors during mode 5 (i.e., most of the mission) for the small FOV. The second curve shows the effective area weighted by e-1.7, the photon spectrum which gives approximately equal counts in all bands. Shafer (1983) discusses the effective areas for the MED and HEDs, as well as the solid angles for the large and small FOV.
Corrections to the Data
All data have been corrected for a slight dependence of observed rate on the observed co-incidence rate (i.e., rejected events). This has been measured on a major frame by major frame basis. All data are binned in 12 hour periods during which the spin axis of the satellite remained approximately stationary and pointed at the Sun. Data from each 12 hour period are filtered for their quality and then cast into 720 spatial intervals, each covering a quarter degree of scan angle (an ecliptic latitude coordinate). Sets of 12 hour stripes were combined for a 180 day period to produce raw maps in ecliptic coordinates. These were then transformed and/or projected in a number of ways to make them more useful.
Table 3: Standard Weights Used to Define X-ray Colors
Color Mode HED Windows MED Windows
The all-sky maps are available in four different forms: raw, ecliptic, celestial, and galactic. Appropriate projections have been used for the celestial and galactic maps. All maps are available in either of the small or large FOV and in each of the four X-ray colors. Ecliptic maps are available in an unprocessed and processed format. We used two methods to process the data: reconstruction of missing data (SAMPLED) and smoothing (BLURRED). Galactic and celestial maps are available only in the SAMPLED and BLURRED formats.
The maps were generated by running a program developed by Keith Jahoda (NASA-GSFC) on the DSDISK database. The program accumulates the scanning data available in the vicinity of a given object's position, and indicates the times when the object was in the FOV. Time is divided into 12 hour allocations. Data are accumulated for 25° of scan angle during times when the source is in the FOV. Data are taken from the MED and the HED3 detectors from either the small or large FOV. The database allows you to collect data summed into 0.25° bins along the scan path for each separate satellite spin axis orientation. The program accumulated 180 days of sequential data, roughly that required to map the entire sky. The output from this program was placed in FITS format (see the raw data file description below) and subsequently transformed into the ecliptic, galactic, and celestial formats as appropriate.
In transforming from one coordinate system to another, we took some care to prevent distortion due to pixellation effects. The transformation programs minimized such distortion by subdividing each pixel into 144 pixelettes, transforming each of them separately, then computing within which new pixel's bounds the pixelette fell. New pixels were weighted according to the total number of pixelettes within their bounds. This preserves the intensity of the sources quite well, but it also causes some minor redistribution of the statistical errors. Because a transformed pixel is a combination of portions of several original pixels, the statistical error in the transformed pixel is no longer independent of its neighbors. Users whose applications are sensitive to such considerations are strongly recommended to work in ecliptic coordinates with the .fitsdata files, where no such effects are present.
In general, the scanning exposure was greatest the first 6 months of the mission. However, the discovery scalar definitions were changed on day 322, resulting in sensitivity changes in several colors. The total rate, for instance, shows a step of a few tenths of a count. For many purposes, it will be more convenient to chose a 6 month period which does not include day 322. The maps are dominated by the cosmic X-ray background which is characterized by a photon index of -1.4 (Marshall et al. 1980). As this spectrum is harder than the one used for band normalization, the apparent cosmic X-ray background falls at day 322. Our start day 322 maps do not include any data prior to this change.
The files in the HEASARC HEAO-1 A2 area on legacy.gsfc.nasa.gov (/FTP/heao1/data/A2/scan/maps) with the filename suffix .fitsdata are raw maps. (Check the online README for updates on directories and filenames.) Strictly speaking, these are not all sky maps, but merely FITS format DSDISK output strips lined up next to each other. Nevertheless, some users may find their applications are sensitive to the kinds of processing we used to make other maps. These files contain data and errors which are statistically independent from pixel to pixel. Standard FITS image viewers (e.g., SAOImage) can be used to view this data, which will appear as an all-sky (or nearly all-sky) image in a ecliptic coordinates. However, the coordinates cannot be read out from the display in quite the normal fashion. Since the satellite traveled with the Earth in an elliptical orbit around the Sun, neighboring rows in the primary array are not necessarily spaced by exactly half a degree in ecliptic longitude, nor is the spacing between adjacent rows perfectly even. The .fitsdata files contains DSDISK data for a 180 day period: for some (e.g., starting day 249), this is sufficient to cover the entire sky, for others (e.g., starting day 429), the Earth's slower motion in its orbit near aphelion results in a 180 day period containing slightly less than 360 degrees of sky coverage.
Since the coordinate spacing is variable, we made a BINTABLE extension to store the coordinates. The BINTABLE contains two columns, ELON and ELAT, which contain the longitude and latitude for the X & Y pixel coordinates, respectively. For instance, pixel (10,20) has a row of ELON and a row of ELAT. We excluded the normal CRPIXn and CDELTn primary array keywords, as they would be both misleading and inaccurate. As of the date of this writing, no further information has been placed in these BINTABLE extensions.
In addition, these raw files also have a second plane in the primary array which contains the errors associated with the pixels in the first plane. This error plane is unique to the raw files; it has not been carried through to other formats and projections. For most applications, the cosmic variations on scales comparable to the FOV dominate the observed scatter.
Given the considerable inconveniences of the raw data, we attempted to make a set of more useful maps which addressed these difficulties while disturbing the data as little as possible. We therefore made a set of FITS image files which contain the raw data in regridded form. Pixels are exactly 0.5° x 0.25° (in a Cartesian projection). When necessary, additional data from adjacent 180 day periods were added to produce an all-sky map: the DATE-OBS and DATE-END keyword values store the range of dates covered by the data. Negative pixel values (from "discovery scalar rollover") and blanks were set to null. The pixel coordinates are stored in the standard FITS keywords. The ecliptic maps include data from day 249 (the first day the MED was stable) until the mission's end.
For some purposes, the presence of null data are irritating. Missing stripes can also be distracting. Therefore we made a set of SAMPLED data in which missing data were "reconstructed" by computing a beam weighted average of all the non-null neighboring pixels. This represents a "best guess" as to what the A2 experiment would have observed had it not been turned off, pointed elsewhere, or had the data not been lost in transmission. Users should consult with the raw maps if they are making intensity measurements from these maps to understand what portion of their measurements may be based on best guesses rather than actual measurement.
Ecliptic maps have also been made in a BLURRED format. In the BLURRED data, we have used a "beam-weighted" average to blur the data to smooth out fine variations which most likely are not real (blur is size of beam: 1.5° x 3.0° or 3.0° x 3.0°). As in the SAMPLED data, we have reconstructed missing data. The blurring makes real fluctuations in the diffuse X-ray background stand out from the instrument noise. However, it also smears sources making them appear more extended than they are, and increases potential source confusion, particularly in the galactic center region where there are a larger number of neighboring A2 X-ray sources. By contrast, the SAMPLED data do not suffer from this extra source confusion.
In simpler terms, SAMPLED data has replaced null pixels with a beam-weighted average of non-null neighboring pixels. BLURRED data has used the same formula, but instead of only replacing null pixels, BLURRED files have replaced ALL pixels with a weighted average of their neighbors. The effect is to fill in missing data in SAMPLED data without affecting real spacecraft data, while BLURRED data has smoothed out fine variations over the entire map.
The galactic maps are available in three different types of projections. There is a simple Cartesian projection, a pair of Aitoffs, and a pair of polar projections. The Aitoffs are centered either on (l,b) = (0,0) or (180,0). The polar projections are zenithial equal area projections (see Greissen and Calabretta (1994) for a definition of the ZEA projection) centered either on the North or South Galactic Pole and extending to |b| = 30°. Thus any source or region of interest should be reasonably well centered in at least one projection.
The galactic maps are available for two date ranges chosen to avoid the most dramatic discovery scalar changes, specifically 322 - 502 and 545 - 628. Coverage is generally better in the earlier epoch as the scanning mode of the A2 instrument was interrupted more frequently later in the mission to perform a variety of pointed observations.
Users already familiar with the Aitoff projections previously available in the legacy FTP area may wish to return and examine the new ones. In addition to having added projections for the anti-galactic direction and the poles, the new Aitoffs have significantly higher spatial resolution. Along the galactic equator, pixels are roughly 0.5° x 0.5°, compared to the 2.0° x 2.0° pixels in the previous Aitoffs.
Celestial maps were made to cover the entire duration of the mission, starting on day 249. As well as a rectangular projection, each celestial map has been made into a set of 50° x 50° tangent plane projections, with approximately 10 degrees of overlap with adjacent maps in both right ascension and declination. Thus any source of interest should be reasonably close to the center of at least one map.
Separating Instrument and Cosmic Background
The A2 detectors employed a novel construction to allow for the instantaneous separation of instrument from cosmic components of the background. The large and small FOV illuminated alternate anodes within a common gas volume. Under the assumption that measured instrument background will be the same throughout the counter while cosmic background will be proportional to the "grasp" (i.e. the area-solid angle product) of the detector , we write (Boldt 1987):
where I is the instrument background rate, C is the cosmic background rate in the small FOV, and are the observed rates in the large and small FOVs, respectively, and [[alpha]] is the ratio of the large field grasp to the small field grasp. The ratio of grasp is nearly the same for the HED3 and MED detectors. Since and are measured, C and I may be determined. In practice, it is often better to determine a mean instrument background for each ecliptic longitude. Recent work with the A2 scanning data (Jahoda and Mushotzky 1989; Jahoda, Lahav, Mushotzky, and Boldt 1992, 1993; Miyaji, Lahav, Jahoda, and Boldt 1994; Boughn and Jahoda 1993) has primarily used the total rate and
One thing to note is that the online unprocessed ecliptic maps show stripes of constant ecliptic longitude (e.g., longitude 199.5 deg in the starting day 545 maps, most obvious in the hard X-ray maps). The stripes appear in both the large and small FOV maps and are believed to be evidence of briefly enhanced instrument background. These disappear if instrument background is calculated (using the above prescription) and removed. The stripes are less obvious in the BLURRED images as these modest enhancements are averaged over many pixels.
What is Planned for the Future?
The three A2 subset databases described above are all in safe keeping, along with the software which accesses them. Conversion of the data into FITS (and revamping the software to deal this, then making the programs available online) will begin in late 1995. This will allow users to examine the data using their own choices of input parameters, for any region of the sky.
Boldt, E. 1987, Phys. Rev, 146(4), 215
Proceed to the next article Return to the previous article
HEASARC Home | Observatories | Archive | Calibration | Software | Tools | Students/Teachers/Public
Last modified: Monday, 19-Jun-2006 11:40:52 EDT
HEASARC support for unencrypted FTP access ended on September 20, 2019. Please see this notice for details.