3.2.1 What about an irregularly shaped object ?

Of course, we do not expect to deal with perfect spherical objects in the Kuiper belt, while we need to model a random occultation with an isotropic diffraction pattern (i.e. which does not show preference to any direction in the observation plane). It is thus important to estimate the effects of an irregularly shaped object on the diffraction pattern. This study has been made by [Roques, Moncuquet and Sicardy 1987] and the Fig.1(bottom) shows the diffraction pattern produced by an elliptical object (e=0.7 and with similar size to the circular case, see caption) occulting a point star, computed from the code defined in that paper. To generate an extreme case, which could a priori smooth the diffraction fringes, the limb of the object has been corrugated by setting the ``irregularity'' parameter n to 3 [Roques, Moncuquet and Sicardy 1987, see for its definition ], that is an object about 6% hilly. Along the two axes of the ellipsoid, the diffraction fringes looks like those of the circular object, while their structures are more complicated outward. But, if we compare the projected contours of the top and bottom of Fig.1, one can see that these fringes are detectable, for a given sensitivity threshold (4% here), at distances larger or roughly equal to the distances where the fringes of the circular object are detected. We think therefore that using a circular object to compute the occultation lightcurve is a good ``working compromise'' for modeling an occultation by a realistic KBO.