of the solar luminosity. Since the observations should be at
least as long as a planet period to detect a transit, searches for
long-period planets similar to those in our solar system require time
consuming observations. In addition, long-period planets have a large
orbital radii, and hence, a weak probability to present the right
line-of-sight inclination of the orbital plane that will produce an
observable transit. Therefore, thousands of stars should be observed for
several years with high precision photometry to exploit the method of
transits. These constraints have seriously limited the applications of
the method after it was first proposed.
Searches for extrasolar planets based on the method of transit are now starting in earnest after the discovery of Jupiter-like planets, with orbital periods much shorter than that of Jupiter, around solar-type stars (Mayor & Queloz 1995; Marcy et al. 1996; Butler et al. 1997; Cochran et al. 1997). A network was organized to search for transit of extrasolar planets (the "TEP'' network). Moreover, space missions have been proposed to conduct long and continuous observations of large samples of stars to exploit the method of transits (Koch et al. 1996; Borucki et al. 1997). One of the main programs of the satellite COROT, that will be launched in 2001, is to achieve this goal (Schneider et al. 1997).
Soon, therefore, we expect to obtain stellar lightcurves showing possible transits of extrasolar planets. Some of these planets will have satellites, that may also transit over the star and be detectable. The expectation rate of planetary transits relies on models of the formation and evolution of planetary systems. Little is known on the distributions in number, size, orbital radius of extrasolar planets. Moreover, the absolute rate of occurence of Jupiter-like planets and the relative occurence of Jupiter-like with respect to Earth-like planets are unknown. Jupiter-like planets should produce stronger transit signals because terrestrial planets are significantly smaller, the expected maximum radius being less than about 2.5 times the Earth radius (Guillot et al. 1996). In this context, the possibility for Jupiter-like planets to have terrestrial satellites raises a crucial issue about the possibility to detect Earth-like objects outside our solar system. A satellite should imprint characteristic signatures on the transit lightcurve of a planet. In addition, the presence of satellites may increase the overall probability of detecting transits of extrasolar planets. The detection of planetary satellites with the method of transits, however, has never been investigated.
In this paper we analyze the probability of detecting satellites of extrasolar planets with the method of transits. Detection rates are computed under the assumption that a planet-satellite system orbits around the target star, and that the duration of the observations is at least as long as the planet orbital period. In Sect. 2 below we compute the probability to detect directly a satellite transit (i.e., the satellite is extended enough to produce a detectable minimum in the lightcurve), separating the cases when the parent planet does and does not also transit over the star. We present the results of this analysis in Sect. 3 and show examples of transit lightcurves revealing the presence of Jupiter-like and terrestrial satellites. Finally, in Sect. 4 we describe the case in which the satellite is too small to produce a detectable minimum in the lightcurve, but its presence may still be inferred indirectly by the timing of the planetary transits.
Copyright The European Southern Observatory (ESO)