3 The Input File

The input file is an ordinary text file and is terminated either by end-of-file or by a record beginning E. Uppercase and lowercase are both acceptable throughout the file and may be mixed freely; leading spaces are ignored. A comment, beginning *, can be appended to any record. Completely blank records (and any beginning with *) are ignored and can be used to improve layout and provide commentary.

Most of the records consist of (or contain) various numbers of numeric fields, separated by spaces (or commas). In many instances it is simply the number of fields present which enables ASTROM to determine which sort of record has been read. Free-format number decoding is used throughout; spaces can be freely inserted between fields, and many other freedoms are permitted (see the documentation for the SLALIB routines SLA_DFLTIN and SLA_DBJIN in SUN/67).

Each file typically specifies a single plate reduction; however, multiple sequences of records, each specifying a complete and separate plate reduction, can be used, each sequence being separated from the next by a record beginning /.

The overall layout of each sequence is as follows:




Group Records Mandatory?



results equinox 1 no
telescope type 1 no
plate data 1 yes
observation data 1-3 no
reference stars 2-3 per star at least 2 stars
unknown stars 1-2 per star no



Several sorts of record involve celestial positions. Although ASTROM can be made to work internally in observed coordinates (i.e. as affected by refraction etc.), all input data and reports are in terms of various sorts of mean [α, δ]. In any particular instance, the mean coordinate system is specified by quoting an equinox. An epoch is also required, for the calculation of proper motion; in the case of catalogue stars this is frequently the same as the equinox.

Both the old pre IAU 1976 (loosely FK4) system and the new post IAU 1976 (loosely FK5) system are supported, and data in the two systems can be mixed freely. ASTROM follows the established convention of using the equinox to distinguish between the two systems. If the equinox is prefixed by B (which stands for Besselian) then the position is an old FK4 one; if a prefix of J is used (standing for Julian), the position is a new FK5 one. If no prefix is used, pre 1984.0 equinoxes indicate the old FK4 system, and equinoxes of 1984.0 or later indicate the new FK5 system. The B or J prefix may also be used with epochs although the distinction is unlikely to be significant. The two most common equinoxes are B1950 and J2000. All FK4 positions include E-terms of aberration, consistent with published catalogues.

When using the old system (e.g. B1950) to specify the position of an object whose proper motion is presumed to be negligible, it is necessary to specify an epoch (as well as the equinox). This is because the old system is not an inertial frame; failure to recognize galactic rotation at the time the system was first established means that the FK4 frame is rotating, and that even extragalactic objects have fictitious proper motions which need to be taken into account in precise work. ASTROM accepts an optional format of reference star data for such cases, where the proper motions are omitted and an epoch is mandatory.

For most ASTROM applications, star positions in the frame of the Hipparcos catalogue can be assumed to be equivalent to FK5 J2000. However, it is unwise to mix Hipparcos and pre-Hipparcos positions. If only Hipparcos-compatible stars are used, and if for ASTROM purposes they are regarded as FK5 J2000, the results of the ASTROM fit will be in the Hipparcos frame to a high degree of accuracy.

Appendix A contains a detailed specification of the syntax of the ASTROM input file, together with a comprehensive example. As an introduction, we will look at a simple but typical example of such a file:


B1950                                     * Results in FK4
SCHM                                      * Schmidt geometry
19 04 00.0  -65 00 00  B1950.0  1974.5    * Plate centre, and epoch
18 56 39.426  -63 25 13.23  -0.0002  -0.036  B1950.0  * Ref 1
44.791   85.643
19 11 53.909  -63 17 57.57   0.0058  -0.044  1950.0   * Ref 2
-46.266   92.337
19 01 13.606  -63 49 14.84   0.0020  -0.026  1950.0   * Ref 3
17.246   64.945
19 08 29.088  -63 57 42.79   0.0016   0.018  1950.0   * Ref 4
-25.314   57.456
19 02 10.088  -63 29 16.73   0.0012  -0.019  1950.0   * Ref 5
11.890   82.766
-5.103    58.868                      *  Candidate
19 09 46.2  -63 51 27  J2000.0        *  Radio pos
END

Taking each record (or group of records) in turn:


B1950                                     * Results in FK4

This is the results equinox record. It specifies the mean equator and equinox for the coordinate system of the report. If this record is omitted, the results will be in J2000.


SCHM                                      * Schmidt geometry

This is the telescope type record, which describes the projection geometry. The telescope type is given by the first four characters of the record; there are currently six options. SCHM, as used here, is for Schmidt telescopes. ASTR selects the tangent plane or gnomonic projection, produced by conventional astrographic telescopes (and by pinhole cameras). Then there are three special AAT options: AAT2 and AAT3 for the Prime Focus doublet and triplet correctors, and AAT8 for the f /8 Ritchey-Chrétien focus (using the vacuum plateholder). The option JKT8 models the field distortion of the JKT (f /8 Harmer-Wynne focus). Finally, the option GENE specifies generalized pincushion/barrel distortion, the magnitude of which is given by a single numeric parameter q following the telescope type string; further details are given in Section 5.


19 04 00.0  -65 00 00  B1950.0  1974.5    * Plate centre, and epoch

This is the mandatory plate data record, specifying the point on the sky corresponding to the pole of the projection geometry (which is usually, but not necessarily2, at the geometrical centre of the plate) and the date on which the exposure occurred. The [α, δ] is expressed as h m s   . The hours, minutes, degrees and arcminutes fields must all be integers. The sign of the δ precedes the degrees. The [α, δ] must be followed by an equinox. The epoch specifies when the picture was taken, needed for the proper motion calculation. The epoch can be omitted if more precise information is to be supplied later via the optional observation data records.


18 56 39.426  -63 25 13.23  -0.0002  -0.036  B1950.0  * Ref 1
44.791   85.643

This is the first of several record pairs describing the reference stars. At least two such pairs are needed in order to run ASTROM, and three if both sorts of linear fit are to be done. At least 10 stars are required if fitting of the radial distortion and/or plate centre is to be attempted. A typical number for an ordinary linear fit would be about a dozen stars; a thorough job covering a large area of a plate and with automatic determination of the radial distortion and plate centre selected would require perhaps 50. The current limit is 2000. The first record contains [α, δ], proper motions in seconds per year and arcseconds per year respectively (n.b. not centuries as in some catalogues), and equinox, followed optionally by epoch and/or parallax. If the epoch is omitted (as in the above example), it is assumed to be the same as the equinox. Parallax, which is in arcseconds, defaults to zero. For reference stars whose positions are given in the old FK4 system, and whose proper motions are presumed to be zero in an inertial sense, an alternative format is available, with the proper motions omitted and the epoch mandatory:


18 56 39.422  -63 25 14.00  B1950.0  1971.3           * Ref 1

(Because the FK4 system is not inertial, using the format described earlier and simply putting zero for the proper motions would not give the same effect.) The first 10 characters of any comment are picked up and appear on the reports as ‘name’. The second record of the reference star pair is the measured [x, y]. For the 4-coefficient model to work properly, x and y must be in the same units. The reports will look best if the units are millimetres or thereabouts and the offsets from zero are reasonably small. Orientation and handedness are immaterial; x=east and y=north is the recommended convention as it matches the run of α and δ.


-5.103    58.868                      *  Candidate
19 09 46.2  -63 51 27  J2000.0        *  Radio pos

These two records both specify unknown stars. The first is [x, y], from which [α, δ] will be determined. The second is [α, δ] and equinox, from which [x, y] will be determined. It is not necessary to include any unknown star records at all, if the intention is simply to determine a plate scale or to check a set of plate measurements.

The above example gives a position for the first unknown star of 19 05 01.794 -63 56 16.70. The equinox was specified in the results equinox record, and the epoch in the plate data record. One way to express this information in a publication might be as follows:

19 05 01.79  -63 56 16.7  B1950  epoch 1974.5

The example does not include the optional observation data and colour records, which enable ASTROM to reconstruct the precise appearance of the field rather than allowing various predictable rotations and distortions to be absorbed into the fit. Also omitted from the example are requests to include the radial distortion and plate centre in the fit. Details of these refinements are given in Sections 7 and 8.

2JKT f /8 plates, for example, are mounted eccentrically. The plate data record must specify the celestial coordinates of the optical axis, preferably to within a millimetre or so. Section 8 gives details of how the plate centre (and the radial distortion) can be determined automatically.