The need to go to space


The science goals of PLATO require the detection of a large number of exoplanets and the detailed analysis of their central stars, including seismic analysis. They are achievable by a very high precision, very long duration and high duty cycle photometric monitoring from space of a very large sample of stars. The Earth's atmosphere provides strong disturbances which limit the achievable performance to millimag accuracies, mostly through scintillation noise. The small amplitude of the photometric dips caused by terrestrial planets are therefore beyond the range of ground-based observations. Scintillation noise in ground-based photometric observations also prevents the detection of low amplitude oscillations in cool stars. The noise in ground-based photometric observations is such that only giant planets with sizes larger than that of Saturn can be detected via their transits, while oscillations of cool stars are totally beyond reach of this technique.

In addition, long, uninterrupted observations, that only space-based instruments can provide, are necessary to optimize the probability of transit detection, and to see several successive transits in order to measure the orbital period, as well as to avoid sidelobes in stellar oscillation power spectra.

Space is therefore necessary on one hand because of its tranquility and the absence of photometric disturbances, and on the other hand because of the possibility it offers to perform the long, uninterrupted observations that are needed to detect exoplanets and to perform seismic analysis of stars.

Alternative techniques can be used from the ground to achieve exoplanet detection as well as oscillation mode detection, and has seen tremendous
progress in recent years. The most efficient of these relies on radial velocity measurements, performed in high resolution spectroscopy, with either echelle cross-dispersed spectrographs (Bouchy et al. 2001) or Fourier transform spectrographs (Mosser et al. 2003). In particular, the HARPS spectrograph installed on the ESO 3.6m telescope at La Silla showed recently that it can reach radial velocity precision as low as about 1 ms-1. Such a low level of radial velocity noise recently led to the discovery of a planet with a mass of 5sini Earth masses on a close-in orbit around the 0.31 solar mass M dwarf Gl 581 (Udry et al. 2007). Although quite impressive, this performance cannot compare with those of a space photometric mission, and would not allow us to reach the science goals of PLATO.

One drawback of the radial velocity technique is that the mass determination suffers from the sini ambiguity, except in the rare cases where the inclination angle i is known. For instance, the recently discovered planet around Gl 581 may well be a much more massive one whose orbit is seen at high inclination.

Second, there is a hard limit for the radial velocity precision achievable with this technique. A Doppler shift of 1 ms-1 corresponds to a spectrum
displacement of about 10 nm on the detector of a HARPS-type spectrograph at 100,000 spectral resolution. The data on which the recent exoplanet discovery around Gl 581 is based are not photon-noise limited, indicating that this hard limit has probably been met. Therefore, there is little hope for improving significantly the performance of the Doppler technique in the future. This means that smaller planets, with sizes and masses comparable to those of the Earth, will never be detected by this technique, and that the exploration of the exoplanet distribution down to at least Earth size, one of the most important science goals of PLATO, is not achievable from the ground.

Observations with HARPS have also revealed solar-like oscillations in nearby bright stars (e.g. Mosser et al. 2005). However these observations have not allowed us to perform full asteroseismic analysis so far, for three reasons:
- the noise level is still too high to detect a large number of oscillation modes;
- the day-night alternance creates strong day-aliases which pollute the power spectra and make their exploitation impossible;
- the total duration of these observations is too short to allow us to measure the oscillation frequencies with a sufficient accuracy.
Besides, asteroseismology with the radial velocity technique is limited to stars with projected rotation velocities smaller than about 10 kms-1 which makes the list of accessible targets very limited.

There is little hope to improve significantly these three major drawbacks of ground-based asteroseismology. The noise level will not be easily decreased further, as discussed earlier. Moreover, these observations will remain severely limited in terms of target magnitude and vsini. Seismology programmes with HARPS are limited to stars with mV about 6, above which photon noise is simply too high to allow us to detect any oscillation mode. In order to reach stars with mV = 11, which is needed to study open cluster members for instance, much larger telescopes with diameters of the order of 40m should be used. High efficiency, high stability, high resolution spectrographs for these extremely large telescopes will be difficult to build, and obtaining high duty cycles for long periods of time on these telescopes is out of question, as detailed below.

The duty cycle of ground-based observations can be improved by multi-site networks of telescopes equipped with appropriate spectrographs; however, even if such an ambitious ground-based network can be setup in the future, the drift of sidereal time limits to only a couple of months the total time during which a high duty cycle can be obtained. An attractive alternative is to perform these observations from Antarctica. However, even in the overly optimistic scenario where an extremely large telescope would be installed in a high quality site in Antarctica, such as Dome C, and made available exclusively to asteroseismology programmes, the total duration of high duty cycle observations would not exceed three months.

We therefore conclude that asteroseismology from the ground with the Doppler technique, which has already been achieved on a few targets with modest duty cycle and total monitoring time, will remain limited to bright stars with low projected rotation velocities, and high duty cycle observations will be very rarely obtained and for limited total amounts of time.