Bureau des longitudes URA 707 CNRS 77, av. Denfert Rochereau F-75014 Paris, France e-mail: berthier@bdl.fr
The observations of occultations of stars by asteroids (and more generally by solar system objects) allow many applications such as highly astrometric measurements ($\approx 0.03 ~arcsec$). It is a good way as well to get planetologic information such as structure, shape, size and neighbourhood of asteroids.
This kind of events occurs many times per day, but there is only a few successful observations each year. The main reason arises from the low precision of predictions and from the small size of the geographic area concerned by these events. The main error $\Delta p$ in the determination of the location on Earth of areas where an occultation is visible are related to the determination of the positions of both the asteroid and the star. An estimation of this error can be defined by the relation $\Delta p = 580 * \rho_{a} \:kms$, where $\rho_{a}$ stands for the geocentric distance of the asteroid in AU, and assuming errors of $0.5"$ and $0.3"$ on the positions of the asteroid and the star respectively.
Hereafter is presented the model developped at the Bureau des longitudes for the prediction of occultations of stars by solar system objects. Precision and possible improvement of the prediction of such events are then discussed, taking into account new stellar catalogues and asteroidal observations.