Transit Data

The Hipparcos Transit Data are described in Volume 1, Section 2.9 of the Hipparcos and Tycho Catalogues (ESA SP-1200, 1997).

The Hipparcos Transit Data file summarises intermediate astrometric and photometric information at the level of individual crossings of targets across the main grid of the Hipparcos instrument. The data is available for 38535 Hipparcos entries selected mainly on account of their known or suspected multiplicity, or as being parts of observations where the signal may be contaminated by the influence of another target close to the edge of the field of view.

Each Transit Data file provides a self-contained and globally calibrated compilation of the main detector signal parameters for almost all transits for specific double or multiple systems. The data can in principle be used to fit arbitrarily complex object models to the Hipparcos observations. For example, the transit data may be used to derive the orbital elements of the resolved components of some short-period binaries with respect to their centre of mass, thus yielding mass ratios in addition to the relative orbits. The data may also be used to obtain astrometric results for some Hipparcos objects that were not successfully solved in time for inclusion in the catalogue, or for which the published results are affected by unresolved ambiguities.

What are the Hipparcos Transit Data?
The Hipparcos Transit Data (TD for short) are a condensed and calibrated version of the raw data from the main detection system onboard the Hipparcos satellite (modulation grid + image dissector tube).  They allow the user to re-interpret the Hipparcos raw data in terms of arbitrary models of the astronomical objects that were observed, without having to worry about complex issues such as the geometric and photometric calibration of the instrument, or how it was pointed as function of time (satellite attitude).  In principle, it should be possible to reproduce the astrometric results published in the Hipparcos Catalogue, including its Double and Multiple Systems Annex, by analysing the TD in terms of the same models [single star, double star with fixed components, etc] as was assumed in the Catalogue, although that would be a rather pointless exercise.  On the other hand, new information can sometimes be generated by assuming a different model, or by injecting supplementary data (e.g. from speckle interferometry) into the solution.

Two important restrictions should be noted however:
  1. TD are only available for about a third of the objects in the Hipparcos Catalogue, or about 38,000 objects.  This restriction was partly to save space (limiting the TD to a single CD-ROM disk), partly because the remaining two thirds are apparently single (unresolved) stars, for which the TD would not add anything significant.
  2. TD are based solely on the results from one of the data reduction consortia, NDAC.  The other consortium, FAST, used the same input (raw) data, but made their own calibrations, attitude determination, etc, which lead to slightly different end results.  In the published Hipparcos Catalogue, the results from the two consortia were compared and combined in order to achieve, statistically, the best possible accuracy on every object.  The TD do not benefit from this combination.  Also, it means that the Hipparcos Catalogue results cannot be reproduced exactly from the TD.
The really raw Hipparcos data (from the main detector) consisted of the individual photon counts obtained with the image dissector tube behind the modulating grid.  1200 such counts (samples) were obtained every second.  The grid produced a modulation with a fundamental frequency of about 140 Hz.  An elementary observation of an object typically consisted of some 200 to 1000 samples spread out over an "observational frame" of 2560 samples (2.133 s) and interlaced with samples of other objects.  These raw samples were affected by all sorts of irregularities of the grid and detector, optical distortion, etc.  By contrast, the TD simply consist of a few Fourier coefficients per observational frame, accurately describing the modulated signal after correction for irregularities, distortion, etc.  While the analysis of the TD is therefore vastly simpler than an analysis of the raw samples, the condensation of the raw samples into the TD was made in such a way that very little information was lost.

Why have the Hipparcos Transit Data been made available?
In the course of the 3.5 year mission, the Hipparcos main detector produced some 13 million elementary observations, such as described above, of some 118,000 different targets.  For the vast majority of the targets, the subsequent data reductions succeeded in determining, accurately and without ambiguity, the astrometric and photometric characteristics as published in the Catalogue and its Annexes.  For a small fraction of the targets (a few thousand) some doubt remained however whether the correct solution was found.  And even if a particular solution is correct, it can sometimes be improved by adding constraints not available at the time of the reductions.  It was therefore desirable to retain a possibility to reconsider these solutions in the future.  This is only possible by going back to the original (raw) data, or at least to some intermediate stage of the data reductions.  The machine-readable version of the Hipparcos Catalogue (on six CD-ROMs) contain three different types of such intermediate data: the Epoch Photometry (see van Leeuwen et al., A&A 323, L61, 1997), the Intermediate Astrometric Data (see van Leeuwen & Evans, A&AS 130, 157, 1998), and the Transit Data.

The Transit Data were thus made available (1) to allow inspection of the Hipparcos data and their interpretation at a more fundamental level than would otherwise be possible; and (2) to allow improved solutions by re-analysis and/or combination with other (space or ground-based) observations.

How can the Hipparcos Transit Data be used?
Two possible ways in which the TD can be used are explained in detail in the paper by Quist & Lindegren (1999): aperture synthesis imaging (producing an "image" of a double or multiple star, from which the number, approximate positions and magnitudes of the components can be inferred), and model fitting (non-linear least-squares fitting of arbitrary object models to the data).

Availability of Data and Software
  1. The Hipparcos Transit Data are available on CD-ROM disk 6 of the Hipparcos and Tycho Catalogues.
  2. To do aperture synthesis imaging, you first need to convert the Hipparcos Transit Data for the selected object into UV-FITS format.  A conversion program in Fortran is available here: td2uv.f.  Then you need some software that can do aperture synthesis imaging.  If you do not already have access to packages such as AIPS, we suggest you try difmap.  Difmap works together with pgplot for visualising and interacting with the images.
  3. To do model fitting, you first need to extract the relevant Hipparcos Transit Data for the selected object.  A Fortran program is available here: td2gf.f.  It produces a simple text file with tab-separated columns of data.  A description of the format is found here.  The format is actually in the form accepted by GaussFit, a general robust least-squares fitting program, but it should be easy to adapt the data to any other model-fitting software you want to use.

Sample files
The paper by Quist & Lindegren (1999) used the double star HIP 97237 as an example both for aperture synthesis imaging and model fitting.  Various input/output files used in that example are available and briefly described below.

Transit Data (in original form):

td097237.td (text file):  this is an exact replica of records 3561383 to 3561635 (inclusive) of the HIP transit data file hip_j.dat on CD-ROM disk 6.  It contains 253 records of 127 bytes (including both CR and LF in the last two bytes).  The first record is the header record, the second is the pointing record, and then follow 251 transit records, one for each passage of HIP 97237 across the main field of view in the Hipparcos instrument.

Files for aperture synthesis modelling:

td097237.uvf (binary file): UV-FITS file used as input for difmap (produced by td2uv)

Files for model fitting, using GaussFit:

mul2_fix.mod (text file):  model file (multiple star with 2 components, fixed relative positions)

td097237.env (text file):  GaussFit environment file
td097237.dat (text file):  data file - input version (produced by td2gf)
td097237.par (text file):  parameter file - input version (starting values)
td097237.res (text file):  result file

Note: the data in the result file are not completely identical to those shown in Fig. 9 of the paper by Quist & Lindegren (1999).  The differences are of the size expected from running td2gf and GaussFit on different computer platforms.

Known Errors in the Hipparcos Transit Data file
Six records in the Hipparcos TD file (ESA SP-1200, CD-ROM disk 6, file cats/hip_j.dat) contain errors caused by format overflow in the Fortran program originally generating the file.  The corresponding fields thus contain asterisks instead of the intended numerical data.  This may cause an error when attempting to read these records.  The programmes td2uv.f and td2gf.f mentioned above, which extract data from this file, automatically correct the erroneous items. This file contains a description of the errors.

References
Example of results obtained after the publication of the Hipparcos and Tycho Catalogues, using the TD:
  • Söderhjelm S., Visual binary orbits and masses post Hipparcos, A&A 341, p.121-140 (1999)
  • Quist C.F., Lindegren L., The Hipparcos Transit Data: What, why and how?, A&AS 138, p. 327-343 (1999)
  • Söderhjelm S., Lindegren L., Perryman M.A.C., Binary star masses from Hipparcos, in ESA SP-402 HIPPARCOS Venice '97, p. 251-256 (1997)
  • Quist C.F., Lindegren L., Söderhjelm S., Using Hipparcos transit data for aperture synthesis imaging, in ESA SP-402 HIPPARCOS Venice '97, p. 257-262 (1997)
Prepared by Lennart Lindegren and Carl Fredrik Quist