next previous
Up: The Hipparcos transit data: how?


3 The transit data files

  This section gives a rather detailed description of the contents of the TD files, complementing the formal description in Vol. 1 of ESA(1997) and providing additional explanation of the data items. All relevant data are contained in two ASCII files, both located on Disk 6 in Vol. 17 of ESA (1997): the TD index file (hip_j.idx, $\simeq 1$ Mb) and the TD file (hip_j.dat, $\simeq 553$ Mb). A summary of the contents of the two files is in Table 1, while important relations among the data are illustrated in Fig. 2.

3.1 The index file: hip_j.idx

 The TD index file is included to facilitate accessing the TD file. It contains a pointer from the HIP number to the corresponding record in the TD file where data on that object can be found. Since HIP numbers range from 1 to 120416, there are exactly 120416 records in the index file. Each record consists of a 7-character integer, which is the record number in the TD file for the relevant header record (Sect. 3.2.1). For example, TD for the double star HIP 7 can be accessed by first reading record number 7 in the index file (hip_j.idx). The content of that record is the integer 129. Record number 129 in the TD file (hip_j.dat) is thus the header record for HIP 7, and information on that object is contained in that and subsequent records of the TD file. If no TD are available for a given HIP number, then the corresponding index file record contains the number -1.

\includegraphics [width=13cm,clip]{ds1699f2.eps}\end{figure} Figure 2: Relations among the contents of the index and data files

For two- and three-pointing systems (Sect. 2.7) there are two or three different index file entries pointing to the same TD record. An example is shown in Fig. 2, where the index file entries for HIP 421 and HIP 424 both point to the 17238th record in the TD file.

For efficient accessing of the TD it is recommended that the index file is read into computer prime memory as a one-dimensional integer array of length 120416. This array is then used as a look-up table for finding data in the TD file. The TD file has a constant record length of 127 bytes (including the two end-of-record characters $\backslash$r$\backslash$n = CR+LF), which allows simple direct access to any given record number.

3.2 The TD file: hip_j.dat


3.2.1 The header record

  The header record is the first record in the TD file containing data on a specific system. The index file always points to this header record for a given HIP number. The header record contains general information about the system in 13 data fields. In holding to the Hipparcos Catalogue conventions, the 13 data fields in the header record will be called JH1, JH2, $\dots$, JH13.

The first three data fields (JH1-3) contain the HIP numbers relevant to the subsequent transit records. JH1 is always defined; JH2 and JH3 are only defined for two- and three-pointing systems (Sect. 2.7). The number of pointings, or more accurately the number of different target positions ($N_{\rm P}$), is specified in field JH4. Details on the relative pointings are given in the pointing record (Sect. 3.2.2).

JH5 gives the number of transits ($N_{\rm T}$), or observations, for a given system. Since there is one record for each transit, this is also the number of transit records. The total number of records for an object is then $N_{\rm T}+2$, where the header and pointing records make up the extra two records.

JH6-JH10 give, in order, the right ascension (deg), declination (deg), parallax (mas), proper motion in right ascension (mas yr-1) and proper motion in declination (mas yr-1) of the reference point. The Fourier coefficients b1-b5 of the subsequent transit records are expressed relative to this reference point. It is important to realize that these values are usually derived from the Input Catalogue and therefore do not agree with the astrometric parameters given in the Hipparcos Catalogue. For instance, the parallax of the reference point is often zero. The position refers to the epoch J1991.25, which was close to mid mission. The reference system is the same as that of the Hipparcos Catalogue, viz. the International Celestial Reference System (ICRS) (Feissel & Mignard 1998).

JH11-JH13 hold the assumed colour index $V\!-\!I$ (mag) for each of the three HIP numbers in JH1 through JH3. This value $V\!-\!I$ was used in rectifying the signals from each transit.

3.2.2 The pointing record

  The pointing record defines the offset of the target position of each subsequent transit with respect to the reference point. The pointing record may define several different target positions as required for multiple-pointing objects. The actual number of different target positions ($N_{\rm P}$) is given in JH4 of the header record. Each target position is specified by three numbers: the first takes the value 1, 2 or 3, depending on which of the HIP numbers in JH1, JH2 or JH3 that the target position refers to. The second and third numbers give the offset in $\alpha$ and $\delta$, respectively, of the target position from the reference point. These last two values are rounded to the nearest arcsecond.

Because every record of the TD file has 125 bytes, this allowed up to nine different target positions to be defined in the pointing record. In reality the maximum $N_{\rm P}$ was 7.

Most objects used a single pointing, and the astrometric values in the Input Catalogue were sufficiently accurate that no updating was required. Moreover, the Input Catalogue values were usually taken as the reference point for the transit data. In this case $N_{\rm P}=1$ and the first and only entry in the pointing record contains the three values: "1 0 0" (cf. the excerpts for HIP 3 and 7 in Fig. 2). The "1" refers to the HIP number in JH1. The "0 0" means that the target position coincided (to the nearest arcsec) with the reference point. Subsequent transit records have "1" in the first field (JT1) indicating that the target position for each transit was as given in the first entry of the pointing record.

In the trivial case of a single, well-centered pointing, as illustrated by HIP 3 and 7 above, the pointing record is really not needed. However, for those objects with multiple HIP entries and pointings, the pointing data provide important information. This is exemplified by the excerpt for the two-pointing system HIP 421+424 shown in the lower part of Fig. 2. As indicated in the header record, there are 341 transits for this system, made using two different target positions. The first transit for this system (in record 17240) was made with the IFOV pointing at the first target position ("2-18-12"), i.e. referring to the second HIP entry (421) and pointing 18 arcsec to the west and 12 arcsec to the south of the reference point. The next transit (record 17241) was made in the second pointing ("1 0 0"), i.e. referring to the first HIP entry (424) and centred on the reference point. The third transit (record 17242) was also made in the second pointing, while the fourth transit was made in the first pointing, and so on.

The fact that Hipparcos had two fields of view, separated by the "basic angle" of $58^\circ$, is largely irrelevant for the use of the TD. There is in fact no flag in the TD telling in which field of view a particular transit occurred.

3.2.3 The transit data record

  The actual scan information is contained in the transit data record. This section will cover the entries contained in these records, as well as some physical interpretations.

The first field (JT1) for each record tells which of the $N_{\rm P}$ (= JH4) target positions was used to observe the transit. The second field (JT2) gives the time of the observation, in years, from the epoch J1991.25. The time is measured in Julian years of exactly 365.25 days. The epoch J1991.25 is equivalent to the Julian date JD 2$\,$448$\,$349.0625 on the Terrestrial Time (TT) scale.

The next three entries (JT3 to JT5) contain the spatial frequencies fx, fy and $f_{\rm p}$ defined as in Sects. 2.4 and 2.5. They are expressed in units of rad rad-1 (radian of modulation phase on the grid per radian on the sky).

The entries JT6 through JT10 contain, in coded form, the five Fourier coefficients b1-b5 defined by Eq. (1). JT6 contains the natural logarithm of b1 ($\ln{b_1}$) and JT7 to JT10 contain the normalized coefficients (b2/b1, b3/b1, b4/b1 and b5/b1). This coding was adopted because b1, being the mean intensity of the signal, is always positive and spanning a wide range of values according to the magnitude of the object; b2 through b5, on the other hand, may have either sign but are always numerically smaller than b1.

The next five entries (JT11-15) contain the natural logarithms of the standard errors ($\sigma_1-\sigma_5$) for each of the Fourier coefficients. JT19, the last entry in this record, is a flag for the standard errors. JT19 is normally set to zero. It is set to "1" if the standard errors could not be computed in the normal way, but were estimated roughly from the standard errors of the other transits. A non-zero value in JT19 is thus a warning that the standard errors in JT11-15 should be treated with caution.

JT16-17 contain two colour correction factors, s1 and s2. As previously mentioned, the TD are rectified signals, i.e. corrected for calibrated variations in the geometric and photometric responses. Among the many factors that have been taken into account in this process are the colour effects with respect to position in the field of view and time variations of the photometric calibrations. These corrections depend on an assumed colour for the object, as given in JH11-13. If it should turn out that the assumed colour for a particular transit was considerably in error, s1 and s2 provide a possibility to correct (approximately) the Fourier coefficients accordingly. See Sect. 2.9.3 in Vol. 1 of the Hipparcos and Tycho Catalogues for details of the correction procedure.

The second to the last entry (JT18) deals with the errors in the attitude determination. The phase determination of each scan was basically limited by photon noise, but the standard errors in JT11-15 also contain a contribution from the along-scan attitude uncertainty. It turned out to be very difficult to treat the attitude errors rigorously in the TD, and as a result these errors were generally underestimated. JT18 contains an estimate of the additional attitude noise ($\sigma_{\rm att}$, in mas) required in order to derive, from the TD, astrometric standard errors that are consistent with the main processing of Hipparcos data.

3.2.4 Known errors in the transit data

  After publication of the Transit Data CD-ROM, six cases of Fortran format overflow errors have been discovered in the file hip_j.dat. No useful data are lost because of the errors, but special measures may be needed to read the corresponding records. The interface programs described in Sect. 6 automatically correct these errors. Further details can be found at the Internet address given in Sect. 6.

next previous
Up: The Hipparcos transit data: how?

Copyright The European Southern Observatory (ESO)