1.2 The Hipparcos Mission

Astrometry at first did not benefit from the revolution brought in astronomy by space observations. This was partly due to its lack of ``appeal'' with respect to other fields of astronomical research, but it is fair to say that the lack of ideas on how one could fully exploit the advantages of space environment for astrometric measurements also played an important role. Finally, in the late 1960s a good idea for an astrometric satellite by Pierre Lacroute finally managed to stand out. At the time the case for accurate parallax determinations was very strong, because they were needed to calibrate the extragalactic distance scale by measuring the distances to nearby stars. The traditional method for ground-based parallax determination made use of photographic plates taken with instruments with very long focal lengths, and thus with very small fields of view. These could only yield a star's parallax relative to an average parallax of a few background stars, whose parallactic displacements were in the same direction but in general of different sizes. An estimate of this background parallax gave an approximation for the correction to be applied to obtain an absolute parallax (see Section 2.1). Such estimates were however difficult and uncertain, and it was not uncommon to see, between independent parallax determinations for the same star, differences much larger than the accuracies indicated for the single determinations. As a consequence, the accuracy of ground-based parallax determinations, painstakingly obtained for a few thousand stars, was at best of the order of 8 mas. To a first approximation, the distance range within which the distances can be determined with an accuracy better than a given thresold is inversely proportional to the accuracy of parallax determinations, and thus the number of objects whose distances are ``well-determined'' increases very rapidly as the error on the parallax decreases. Therefore, the poor accuracy achievable from the ground was a severe limitation, for many types of stars of great astrophysical importance, including the most reliable standard candles, were too rare to be found within the surveyable volume. Lacroute's basic idea was to combine the images from two areas widely separated on the sky onto one detector, and to do this several times a year. The parallactic displacements of the two fields would be uncorrelated and therefore allow a reliable determination of absolute parallaxes. This technique could not be employed successfully from the ground, as it required a very well determined and stable angle between the two viewing directions, and determination of large angles from the ground is seriously affected by atmospheric refraction. The first proposal of a space mission implementing this technique was submitted to the French Space Agency in 1966, but its realization appeared too complex at that time, and in 1970 further studies were stopped. The basic idea survived within ESA, however, and thanks to the input of several new ideas originating from the study group assembled by ESA in 1975, the concept of a space astrometry mission became more and more realistic. In 1980, the Hipparcos^1.5mission ([ESA 1997a], [ESA 1997b], [van Leeuwen 1997] and [Kovalevsky 1998]), the first dedicated astrometric satellite mission ever, was finally approved by ESA. The basic requirement in order to have a sufficient scientific impact with respect to ground-based data was then considered the acquisition of astrometric data for about 100000 stars evenly distributed over the sky with an accuracy better than 2 mas for positions and parallaxes and 2 mas/yr for proper motions. The astrometric data would be complemented by multi-epoch photometric data for all stars in the very wide $Hp$ band. A second experiment, proposed by Erik Høg and consisting in an astrometric survey of a million stars complete to about $V=11$, was later incorporated in the mission and named Tycho^1.6. This parallel experiment would also acquire photometric data in two wide bands, $B_T$ and $V_T$, roughly resembling the Johnson $B$ and $V$ bands, that would be useful to the main mission as well in order to correct the observations for chromatic aberration.

The mission design that was established in order to meet these requirements considered a spinning satellite in geostationary orbit whose spin axis preceded around the Sun-satellite direction, or Sun axis, thus allowing a complete and uniform coverage of the sky. The two viewing directions, perpendicular to the spin axis and separated by an angle very close to 58 deg which is called the basic angle, were projected on the same detector by the beam combiner, a mirror that was cut in two halves and glued back at an angle equal to half the basic angle. In order to accurately calibrate the basic angle on the basis of the satellite's smooth rotation, the satellite in general and the beam combiner in particular required a mechanically and thermally very stable environment, which was to be achieved thanks to the large distance from Earth characterizing geostationary orbits and to the constancy at 43 deg of the Sun angle, i.e. the angle between the spin axis and the Sun axis. The images from the two fields of view were then superposed on the focal plane by an all-reflective Schmidt-like telescope working in the visible with $\lambda/60$ rms mirrors and a small aperture of 29 cm. The detector system consisted of photoelectric detectors in conjunction with slit systems for signal modulation, and was in many ways similar to that which is used in meridian circles. In both cases the objects crossed the field of view, as a result of the satellite's and Earth's rotation, respectively. The central part of the focal plane was equipped with an image dissector tube and used for the main experiment, whereas at the two sides two photomultipliers were used for the star detection process that was necessary for the main experiment and for the Tycho experiment.

This revolutionary mission concept, however, posed some challenging problems that had never occurred before. At a very early stage of the mission, it became clear that the reduction of the large and complicated dataset which was to result for any star from the whole mission could not be accomplished in just one run, due to the huge computer resources that would be needed to do so. However, in 1976 already Lennart Lindegren designed a three-step reduction system which was to be adopted and which allowed to handle the problem with only a small loss of accuracy. Even so, the data reduction was to be very delicate, as there was no independent material available with which to compare Hipparcos results, and it was decided that it would be beneficial to the project if more than one group carried out the full data reduction process. This led to the establishment of two consortia, whose collaboration greatly improved the quality of the final catalogues. It was then necessary to select the stars to be observed, on the basis both of the desired scientific goals and of the mission's constraints, the most stringent of which being that the stars had to be uniformly distributed over the sky and that their positions at the epoch of observation had to be known a priori with an accuracy better than $1.5~\textrm{arcsec}$. Following the submission of observing proposals, a laborious selection process involving massive simulation of observations brought to the identification of an optimal observing list of about 120000 stars. All that was known about these in terms of astrometry, photometry, multiplicity information, spectral type and radial velocity was gathered in order to optimize the observation strategy and was later published in The Hipparcos Input Catalogue ([Turon et al. 1992]).

The satellite was designed and constructed by a European industrial consortium led by Matra Marconi Space and Alenia Spazio and on August 8, 1989 was launched into a geostationary transfer orbit by an Ariane 4 launcher. Despite the failure of the Apogee Boost Motor, which was to put the spacecraft into geostationary orbit but left it stuck in a highly elliptical one, the satellite acquired data for 3.5 years, significantly exceeding all the mission goals. In 1997, after painstaking verifications of the data quality and four years after the end of data acquisition, the mission results were finally published by ESA. The Hipparcos and Tycho Catalogues ([ESA 1997a]) contain high-quality astrometric data (positions, proper motions and parallaxes) with an accuracy^1.7in the 0.5-2 mas range for about 120000 stars (Hipparcos), and in the 10-60 mas range for over 1000000 stars (Tycho). The astrometric data were complemented by photometric data in the $Hp$ band for Hipparcos stars and in the $B_T$ and $V_T$ bands for Tycho stars, and a great deal of additional data on variable and multiple stars was made available as well. The final results were about 1.5 to 2 times better than the original aims and improved the accuracy of orders of magnitude with respect to ground-based measurements.

The astrophysical significance of such a large and uniform dataset became apparent early during the mission, and was later confirmed by the high and steady rate of Hipparcos-based articles' publication^1.8.

Besides, despite the careful reduction of the observations that had been carried out, it became clear that important information was still ``hidden'' in the raw Hipparcos and Tycho data. Therefore, even after the catalogues' publication, great care has been given at fully exploiting the scientific potential of the data. As a first step, proper motions with an accuracy of about 2.5 mas/yr were obtained independently by [Urban, Corbin and Wycoff 1998] and [Høg et al. 1998a] for most Tycho stars by comparing Tycho positions with the one-century old positions of the Astrographic Catalogue stars obtained by [Urban et al. 1998]. The recently completed thorough re-reduction of Tycho raw data in combination with a large number of ground-based catalogues has then allowed to increase by a factor of 2.5 the number of observed stars. The resulting Tycho 2 Catalogue ([Høg et al. 2000a] and [Høg et al. 2000b]) contains positions and proper motions for the brightest 2.5 million stars in the sky, complemented by components of 7500 resolved double stars with separation down to 0.8 arcsec. Further dedicated data reduction procedures are now being developed for an optimal treatment of double and variable stars, which will result in the largest duplicity and variability survey ever, the so called Tycho 3 Project. These improvements demonstrate a fundamental principle of astrometric measurements, namely that they are a particularly valuable resource for future analysis and comparison. The content of the Hipparcos, Tycho and Tycho 2 Catalogues are summarized in Table 1.1. These must be considered as the present state-of-the-art in astrometry, and may be compared with the measurement capabilities expected from GAIA, and described in Section 2.7, in order to evaluate the achievable progress.

Table 1.1: Basic results of the Hipparcos mission. The Hipparcos, Tycho and Tycho 2 Catalogues: number of objects, astrometric and photometric accuracy, multiplicity and variability information from [ESA 1997a] (Hipparcos and Tycho) and [Høg et al. 2000a] (Tycho 2).

Hipparcos Catalogue
Entries	118218
Mean Star Density	$\simeq$ 3 stars/deg$^2$
Limiting Magnitude	$V\simeq$ 12.4 mag
Completeness	$V\simeq$ 7.3-9.0 mag
Median Precision of Positions (J1991.25) for $Hp<9$	0.77 mas in RA
	0.64 mas in Dec
Median Precision of Proper Motions for $Hp<9$	0.88 mas/yr in RA
	0.74 mas/yr in Dec
Median Precision of Parallaxes for $Hp<9$	0.97 mas
Parallaxes determinaed to better than 10%	20853
Parallaxes determinaed to better than 20%	49399
Systematic Errors in Astrometry	$<$ 0.1 mas
Solved Double or Multiple Systems	12195 (2996 new)
Median Photometric Precision in $Hp$ for $Hp<9$	0.0015 mag
Periodic Variables	2712 (970 new)
Tycho Catalogue
Entries	1058332
Mean Star Density	$\simeq$ 25 stars/deg$^2$
Limiting Magnitude	$V_T\simeq$ 11.5 mag
Completeness	$V_T\simeq$ 10.5 mag
Median Precision of Positions (J1991.25) for $V_T<9$ mag	7 mas
Median Precision of Positions (J1991.25) for All Stars	25 mas
Systematic Errors in Astrometry	$<$ 1 mas
Median Photometric Precision for $V_T<9$ mag	0.014 mag in $B_T$
	0.012 mag in $V_T$
	0.019 mag in $B_T-V_T$
Median Photometric Precision for All Stars	0.07 mag in $B_T$
	0.06 mag in $V_T$
	0.10 mag in $B_T-V_T$
Tycho-2 Catalogue
Entries	2529913
Mean Star Density	$\simeq$ 60 stars/deg$^2$
Limiting Magnitude	$V\simeq$ 12.4 mag
Completeness to 99%	$V\simeq$ 11.0 mag
Completeness to 90%	$V\simeq$ 11.5 mag
Median Precision of Positions (J2000.0) for $Hp<9$	7 mas
Median Precision of Positions (J2000.0) for All Stars	60 mas
Median Precision of Proper Motions for All Stars	2.5 mas/yr
Systematic Errors in Positions	$<$ 0.1 mas
Systematic Errors in Proper Motions	$<$ 0.5 mas/yr

In summary, the Hipparcos mission put an end to the long-standing strict separation between astrometry and astrophysics by dramatically demonstrating the deep astrophysical significance of extremely accurate global astrometric observations. A further improvement of orders of magnitude in terms of accuracy and number of objects is now expected from the next generation of astrometric satellite missions such as GAIA, proposed to ESA in the framework of its Horizons 2000 long-term scientific programme. These outstanding observational capabilities could be the key towards the solution of one of the major problems in contemporary astronomy, namely the understanding of the structure, formation and evolution of the Galaxy.