As calibration gets under way, the focus in our planning shifts to writing the software to analyze flight data. As a user or archival researcher, you will receive a set of standard data products in FITS, and we are providing a set of data analysis programs to help you work with the data.
The data products will be similar to the usual ones from ROSAT and ASCA. The primary dataset for each instrument is an event list with photon arrival time, pulse height and spatial coordinates. We will include aspect corrected sky coordinates, tangent plane coordinates (telescope off axis angle and azimuth) and chip pixel coordinates. Since our chips are tilted with respect to the focal plane, the tangent plane and chip systems are a little different, which is a new wrinkle. As well as the event list, we provide images in several predefined standard bands, together with matching exposure maps and background estimates, and housekeeping and aspect solution information. These so-called Level 1 products form the basic calibrated science information. Our data processing pipelines will also run automated source detection to generate a list of candidate X-ray sources, and derive a set of Level 2 products which give basic information on each source (extracted pulse height spectra, light curves, and so on). These are provided as a way to take a quick look at the data, but most users will want to reextract the source data themselves. We've decided not to include automated spectral fits and other model-dependent processing, since our experience and that of the Users Committee is that such products haven't been scientifically useful for past missions.
The basic analysis system will inherit a lot of familiar features from the IRAF/PROS (Post-Reduction Off-Line Software) software developed for ROSAT. In particular, you'll have the ability to specify `virtual files' which filter event list data on the fly to select particular regions, time intervals or pulse height ranges without having to make an intermediate file. We will make some enhancements to the PROS filtering syntax so that it's easier to filter on regions and to make images in detector coordinates or in time-energy space. We also intend to write our software to be as generic as possible, emphasizing flexible ways of filtering and binning the data on any science or housekeeping attribute.
A big difference from IRAF/PROS is that you will be able to work in the FTOOLS (a suite of X-ray analysis software) style, at the operating system command line. The programs will also be able to directly read and write FITS files as well as PROS-style QPOE (Quick Position-Ordered Event) files, so you can mix and match our software and external programs such as FTOOLS. You won't need any kind of IRAF installation to use the tools, although if you do have IRAF the tools can be installed as IRAF tasks. This `open' approach reflects our desire to let you work in whichever environment you prefer. FTOOLS and PROS both use an IRAF-style parameter interface, so there's already a lot in common. Some users have asked us about direct support for IDL; we're looking at this, but the IDL paradigm is sufficiently different that we can't promise to provide this support given our resources.
Another major innovation (discussed in an accompanying article) is the use of graphical user interfaces (GUI) - the AXAF system will be one of the first generation of astronomical data systems to incorporate GUIs in its initial design. You'll probably want to use the GUIs for initial data reduction and use command line programs or scripts for repetitive tasks.
The analysis system includes all of the usual X-ray and generic analysis tools, including routines to extract spectra and handle exposure maps. The aspect solution is of particular interest for AXAF, and a variety of extra intermediate aspect solution data products will be available on request, as well as the aspect solution software itself.
The unprecedented spatial resolution available with ACIS and HRC makes spatial analysis and modeling crucial. Many sources which were pointlike with ROSAT will be extended when seen by AXAF. We'll provide a variety of source detection methods, including the standard sliding box and newer wavelet based techniques. We are also developing a spatial modeling tool which will include the ability to fit the energy dependence of spatial effects. You can do standard point source spectral analysis with the usual XSPEC (an X-ray spectral fitting package) program, and we will provide programs to tune response matrices for individual sources.
Grating analysis presents X-ray astronomers with a new set of challenges. The dispersed spectrum will consist of several overlapping orders. When the High Energy Transmission Grating is used in conjunction with the ACIS CCDs, the intrinsic pulse height energy resolution of the CCDs lets you determine the photon order with adequate accuracy. We'll provide tools which can plot the events in pulse height vs. dispersion coordinate space and lay down suitable mask regions to select a given order. Other programs will let you measure and fit functional forms to spectral features, and sum contributions from different orders.
Finally, we think it's important that it's easy for users to write their own programs to analyse the data. At its simplest level, that means easy ways to get the data out of the data files and into your program. The object-oriented generic data model which we've been designing to support generic filtering lets us give you a subroutine interface ( in Fortran and C ) with what we hope you'll find are simple and flexible data access routines that you can use in your software by linking to a set of libraries.
The analysis system is scheduled for release to users shortly before launch; a subset of proposal planning tools will be available in 1997.
Jonathan McDowell