1.1 The Advantages of Space Astronomy

Since the 1950s, artificial satellites have been used for a wide range of applications, thus providing the vast knowledge that was necessary in order to launch and operate relatively low-cost satellite-borne astronomical telescopes. Even though such early missions were primarily motivated by the need of extending the observations to almost all wawelengths of the electromagnetic spectrum, in the following we will concentrate our attention on optical telescopes, owing to the fact that GAIA will observe in the optical region.

Roughly speaking, the performance of an imaging telescope can be expressed in terms of its sensitivity and angular resolution. In principle, both can be improved by increasing the aperture, and thus the light-gathering area^1.2. Unfortunately, in recent years conventional monolithic mirrors have virtually approached the practical maximum size. Besides, however large its aperture may be, the angular resolution of a ground-based telescope ultimately becomes limited by problems connected with terrestrial environment only. In order to cope with these difficulties, a wide range of new techniques, such as mosaic and light-weight mirrors, active and adaptive optics, was developed. Still the most direct way to do so is to position a telescope in a far enough place so as to make the influence of terrestrial environment negligible. This remarkably difficult undertaking is justified by the manifold substantial advantages of space-based astronomy over ground-based astronomy, which can be summarized as follows:

• Atmospheric Turbulence and Refraction: space and time variations in chemical and physical properties of the atmosphere cause variations in the refractive index of different atmospheric regions. Since a telescope collects light over a large area, differences in refraction along different paths lead to random and systematic deviations of light rays, which are referred to as atmospheric turbulence and atmospheric refraction, respectively. The problems posed by these two phenomena are quite different. Atmospheric turbulence smear the image of an otherwise point-like source to a spot whose Full-Width at Half-Maximum (FWHM) is called seeing. The typical seeing at a good observing site is of order 1 arcsec, and only 5% of the observations at a superb site such as Cerro Paranal, where ESO Very Large Telescope is being built, have a seeing better than 0.4 arcsec. Since the diameter of the Airy Disk of an 8 m aperture telescope is of order 35 mas^1.3, Atmospheric turbulence degrades the possible diffraction-limited angular resolution of such a telescope by at least an order of magnitude, and is therefore the main factor limitating the performance of currently available large ground-based telescopes. atmospheric refraction causes a systematic displacement of the positions of celestial bodies of order 1 arcsec, and therefore can in principle be corrected for, but the space and time varying correction to be applied is never perfectly determined, especially at large zenith angles, where the path followed by light rays through the atmosphere is particularly long. This is a strong limitation to the accuracy of wide-angle astrometry, which requires observations carried out with different telescopes and at different times to be compared and combined.
• Atmospheric Extinction: photons reaching Earth from space are scattered and absorbed by molecules and dust that populate the atmosphere. Photons in most wavelength regions are virtually completely absorbed, leaving only two relatively small ``windows'' available for ground-based astronomical observations, namely the optical window ( $300~\textrm{nm}\lesssim\lambda\lesssim800~\textrm{nm}$) and the radio window ( $1~\textrm{mm}\lesssim\lambda\lesssim20~\textrm{m}$). Even in the optical window, however, absorption and scattering dim and redden the celestial bodies to an extent increasing with zenith angle. Since the actual amount of atmospheric extinction is difficult to determine, with a ground-based telescope one usually has to measure the brightness of stars with respect to some reference star whose ``true'' brightness (i.e. the brightness as seen from outside the atmosphere) is reasonably well known, and then deduce the unknown star's ``true'' brightness. This procedure can prove very accurate, but requires that the reference star and the unknown star are very close in the sky, so that one has still to be able to correct for atmospheric extinction the brightness of an all-sky dense net of reference stars.
• Mechanical Flexure: the Earth's gravitational field tends to bend the truss structure of large, heavy ground-based telescopes, so that the resulting image is distorted. This distortion can be made negligible by positioning a satellite far from Earth, e.g. in a geostationary orbit or in the L2 Lagrangian point of the Sun-Earth system.
• Thermal Stability: a ground-based telescope is subject to temperature variations characterizing the Earth's surface, which in turn cause continuous expansion and compression of the instrument's parts. The resulting image distortion can be reduced by putting a satellite in an orbit such that the exposition to sunlight is approximately constant. Besides, the low temperature of space environment simplifies the detectors' cooling, which is necessary in order to reduce the detectors' readnoise.
• All-Sky Coverage: a ground-based telescope can efficiently observe a limited portion of the sky depending on its geographical coordinates. In order to compile an all-sky astrometric catalogue, for instance, one has to combine several partially overlapping regional catalogues. Each of these will introduce its own systematic errors, thus degrading the goodness of the single catalogues' results. For this reason ground-based narrow-field astrometry is usually much more accurate than ground-based global astrometry, where systematic errors often dominate the error budget. This issue is particularly important because any absolute astronomical reference system obtained from radio observations of extragalactic objects, such as the recently established International Celestial Reference System (ICRS, [Arias et al. 1995]), needs a routinely accessible optical counterpart. Hipparcos observations have for instance been linked to the ICRS, thus discovering errors of several tenths of arcseconds in the most accurate optical astrometric catalogue obtained from ground-based observations, the FK5.
• Sky Background: even at a superb observing site, even in the best environmental conditions, the sky is never completely dark. This is a particularly serious drawback when, as in surface photometry, one has to subtract the contribution from the sky background from the observed surface brightness distribution of faint diffuse objects. The sky background as seen from space is substantially fainter than from the ground. Besides, its most rapidly varying components seen in ground-based images originate in the atmosphere and are therefore absent in space observations. This allows a much more accurate correction for the sky background than it is generally possible to accomplish from the ground.

Space observations were first carried out from rockets in the 1960s, but it was not until the 1970s that they were extensively carried out from Earth satellites. Since then, thanks to the many advantages we mentioned, space-based astronomy has continuosly gained in importance. Among other factors, this rapid development has been made possible by the close collaboration between the scientific community, the industry and the new-born national and international space agencies. In a European context, the efforts of the different national space agencies were coordinated by the European Space Agency (ESA), whose activity has in time greatly contributed to the success of the European astronomical community^1.4.