3 Overall design considerations

Instrument design converges through a consideration of technical feasibility and scientific requirements. The proposed GAIA design has arisen from requirements on astrometric precision (10 $\mu$as at 15 mag), completeness to V=20 mag, the acquisition of radial velocities, the provision of accurate multi-colour photometry for astrophysical diagnostics, and the need for on-board object detection (Mignard 1999; Gilmore et al. 2000).

3.1 Astrometry

A space astrometry mission has a unique capability to perform global measurements, such that positions, and changes in positions caused by proper motion and parallax, are determined in a reference system consistently defined over the whole sky, for very large numbers of objects. Hipparcos demonstrated that this can be achieved with milliarcsecond accuracy by means of a continuously scanning satellite which observes two directions simultaneously. With current technology this same principle can be applied with a gain of a factor of more than 100 improvement in accuracy, a factor 1000 improvement in limiting magnitude, and a factor of $10\,000$ in the numbers of stars observed.

Measurements conducted by a continuously scanning satellite are optimally efficient, with each photon acquired during a scan contributing to the precision of the resulting astrometric parameters. The over-riding benefit of global astrometry using a scanning satellite is however not efficiency but reliability: an accurate instrument calibration is performed naturally, while the interconnection of observations over the celestial sphere provides the rigidity and reference system, immediately connected to an extragalactic reference system, and a realistic determination of the standard errors of the astrometric parameters. Two individual viewing directions with a wide separation is the fundamental pre-requisite of the payload, since this leads to the determination of absolute trigonometric parallaxes, and absolute distances, exploiting the method implemented for the first time in the Hipparcos mission.

The ultimate accuracy with which the direction to a point source of light can be determined is set by the dual nature of electromagnetic radiation, namely as waves (causing diffraction) and particles (causing a finite signal-to-noise ratio in the detection process). For wavelength $\lambda$ and telescope aperture D the characteristic angular size of the diffraction pattern image is of order $\lambda/D$radians. If a total of N detected photons are available for localizing the image, then the theoretically achievable angular accuracy will be of order $(\lambda/D)\times N^{-1/2}$ radians. A realistic size for non-deployable space instruments is of order 2 m. Operating in visible light ( $\lambda \sim 0.5~\mu$m) then gives diffraction features of order $\lambda/D \sim 0.05$ arcsec. To achieve a final astrometric accuracy of 10 $\mu$as it is therefore necessary that the diffraction features are localised to within 1/5000 of their characteristic size. Thus, some 25 million detected photons are needed to overcome the statistical noise, although extreme care will be needed to achieve such precision in practice. The requirement on the number of photons can be satisfied for objects around 15 mag with reasonable assumptions on collecting area and bandwidth. Quantifying the tradeoff between dilute versus filled apertures, allowing for attainable focal lengths, attainable pixel sizes, component alignment and stability, and data rates, has clearly pointed in the direction of a moderately large filled aperture (as apposed to an interferometric design) as the optical system of choice.

The GAIA performance target is 10 $\mu$as at 15 mag. Restricting GAIA to a limiting magnitude of 15 mag, or to a subset of all objects down to its detection limit, would provide a reduction in the down-link telemetry rate, but little or no change in the other main aspects of the payload design. These are driven simply by the photon noise budget required to reach a 10 $\mu$as accuracy at 15 mag. The faint magnitude limit, the ability to meet the adopted scientific case, and the number of target objects follow from the accuracy requirement, with no additional spacecraft cost.

3.2 Radial velocity measurements

There is one dominant scientific requirement, as well as two additional scientific motivations, for the acquisition of radial velocities with GAIA: (i) astrometric measurements supply only two components of the space motion of the target stars: the third component, radial velocity, is directed along the line of sight, but is nevertheless essential for dynamical studies; (ii) measurement of the radial velocity at a number of epochs is a powerful method for detecting and characterising binary systems; (iii) at the GAIA accuracy levels, "perspective acceleration'' is at the same time both a complication and an important observable quantity. If the distance between an object and observer changes with time due to a radial component of motion, a constant transverse velocity is observed as a varying transverse angular motion, the perspective acceleration. Although the effect is generally small, some hundreds of thousands of high-velocity stars will have systematic distance errors if the radial velocities are unknown.

On-board acquisition of radial velocities with GAIA is not only feasible, but is relatively simple, is scientifically necessary, and cannot be readily provided in any other way. In terms of accuracy requirements, faint and bright magnitude regimes can be distinguished. The faint targets will mostly be distant stars, which will be of interest as tracers of Galactic dynamics. The uncertainty in the tangential component of their space motion will be dominated by the error in the parallax. Hence a radial velocity accuracy of $\simeq$5 km s^-1 is sufficient for statistical purposes. Stars with  mag will be of individual interest, and the radial velocity will be useful also as an indicator of multiplicity and for the determination of perspective acceleration. The radial velocities will be determined by digital cross-correlation between an observed spectrum and an appropriate template. The present design allows (for red Population I stars of any luminosity class) determination of radial velocities to $\sigma_v \simeq 5$ km s^-1 at V=18 mag (e.g. Munari 1999a).

Most stars are intrinsically red, and made even redder by interstellar absorption. Thus, a red spectral region is to be preferred for the GAIA spectrograph. To maximize the radial velocity signal even for metal-poor stars, strong, saturated lines are desirable. Specific studies, and ground-based experience, show that the Ca II triplet near 860 nm is optimal for radial velocity determination in the greatest number of stellar types.

Ground-based radial velocity surveys are approaching the one million-object level. That experience shows the cost and complexity of determining some hundreds of millions of radial velocities is impractical. There is also a substantial additional scientific return in acquiring a large number of measurements, and doing so not only well spaced in time but also, preferably, simultaneously with the astrometric measurements (e.g. variables and multiple systems).

3.3 Derivation of astrophysical parameters

The GAIA core science case requires measurement of luminosity, effective temperature, mass, age and composition, in addition to distance and velocity, to optimise understanding of the stellar populations in the Galaxy and its nearest neighbours. The quantities complementary to the kinematics can be derived from the spectral energy distribution of the stars by multi-band photometry and spectroscopy. Acquisition of this astrophysical information is an essential part of the GAIA payload. A broad-band magnitude, and its time dependence, will be obtained from the primary mission data, allowing both astrophysical analyses and the critical corrections for residual system chromaticity. For the brighter stars, the radial.

Figure 1: Filter transmission curves and CCD response curves for the provisional (baseline) broad-band (left) and medium-band (right) photometric systems

For essentially every application of the GAIA astrometric data, high-quality photometric data will be crucial, in providing the basic tools for classifying stars across the entire HR diagram, as well as in identifying specific and peculiar objects (e.g. Straizys 1999). Photometry must determine (i) temperature and reddening at least for OBA stars and (ii) effective temperatures and abundances for late-type giants and dwarfs. To be able to reconstruct Galactic formation history the distribution function of stellar abundances must be determined to $\sim$0.2 dex, while effective temperatures must be determined to $\sim$200 K. Separate determination of the abundance of Fe and $\alpha$-elements (at the same accuracy level) will be desirable for mapping Galactic chemical evolution. These requirements translate into a magnitude accuracy of $\simeq$0.02 mag for each colour index.

Many photometric systems exist, but none is necessarily optimal for space implementation. For GAIA, photometry will be required for quasar and galaxy photometry, Solar System object classification, etc. Considerable effort has therefore been devoted to the design of an optimum filter system for GAIA (e.g. Høg et al. 1999a; Munari 1999b). The result of this effort is a baseline system, with four broad and eleven medium passbands, covering the near ultraviolet to the CCD red limit. The filters are summarised in Fig. 1. The 4 broad-band filters are implemented within the astrometric fields, and therefore yield photometry at the same angular resolution (also relevant for chromatic correction), while the 11 medium-band filters are implemented within the spectrometric telescope. Both target magnitude limits of 20 mag, as for the astrometric measurements.

3.4 On-board detection

Clear definition and understanding of the selection function used to decide which targets to observe is a crucial scientific issue, strongly driving the final scientific output of the mission. The optimum selection function, and that adopted, is to detect every target above some practical signal level on-board as it enters the focal plane. This has the advantage that the detection will be carried out in the same wave-band, and at the same angular resolution, as the final observations. The focal plane data on all objects down to about 20 mag can then be read out and telemetered to ground within system capabilities. All objects, including Solar System objects, variable objects, supernovae, and microlensed sources, are detected using this "astrometric sky mapper'', described in further detail in Sect. 4.3.

Figure 2: The payload includes two identical astrometric instruments (labelled ASTRO-1 and ASTRO-2) separated by the 106$^\circ$ basic angle, as well as a spectrometric instrument (comprising a radial velocity measurement instrument and a medium-band photometer) which share the focal plane of a third viewing direction. All telescopes are accommodated on a common optical bench of the same material, and a basic angle monitoring device tracks any variations in the relative viewing directions of the astrometric fields