3 VLT/NAOS simulations

The relatively large pixel size together with the modest spatial resolution of the ESO 3.6 m telescope strongly limits the deconvolution process. The marginal high-resolution that can be achieved with the present data preferably yields a binary system, but conversely we know that such a result can also be obtained if the object is dumbbell shaped. Larger telescopes with a 15-25 mas/pix scale could overcome this limitation. To illustrate this claim, we have constructed simulated images from the NAOS/CONICA camera (with a 13.25 mas/pixel scale) installed on the ESO VLT/UT telescope at Paranal (Chile). Four cases are analyzed corresponding to the dumbbell shape of Tanga et al. (2001), the two binary models obtained (A) from the restored images and (B) from the Roche ellipsoids model described in Sect. 2, and the "bone-shape" model of Ostro et al. (2000). The images are obtained by convolution of these shape models with a simulated H-band (1.64 $\mu$m) NAOS/CONICA PSF profile under an atmospheric seeing which varied between approximately 0.6 and 1.0 arcsec (Rousset et al. 1998). The set of PSF indicates a highly stable correction of the AO system with a Strehl ratio between 56 and 80% and a FWHM of $\sim$31.5 mas. Additional noise corresponding to photon noise and the readout noise of the detector (with $\sigma=120~$e-) have been included in these convolved images. Results from the restoration process with the MLR and MISTRAL techniques are shown in Fig. 2. We stress that the PSF applied for the construction of the simulated image (second row of the figure) is not used during the myopic deconvolution. Instead, we used a mean PSF estimated from several PSFs taken at different seeing conditions for which we can also estimate the variability per element of frequency in Fourier space (the Density Spectral Point or DSP). Comparison of the original model (first row of the figure) with the restored object (last three rows of the figure) shows the advantage of the MISTRAL over the classical MLR method. Moreover, it appears that VLT/NAOS/CONICA observations of (216) Kleopatra would be of great value since they could conclusively reject an incorrect shape-model. The same procedure was applied with appropriate scaling for the apparent size and signal-to-noise ratio to an asteroid similar to (624) Hektor (V=14 and about half the apparent size of Kleopatra, see Fig. 3). Here the H-band PSF is simulated for a fainter object, and the correction is not as good as for Kleopatra. For instance, under a seeing of 0.7 arcsec the PSF nominal Strehl ratio is 31% and its FWHM is 34 mas, but the variation with the seeing quality is much larger (we have SR= 14% and FWHM = 50 mas under a seeing of 0.9 arcsec). Even though the restitution is degraded in this case, the MISTRAL deconvolution method linked with the intrinsic resolution of the 8 m size telescope still enables us to reveal the binary nature of close pairs among the Trojans.