HR: 1340h
AN: P53A-1457    [Abstracts]
TI: Cassini Radio Occultation of Saturn's Rings: a Bayesian Approach to Particle Size Distribution Recovery
AU: * Wong, K K
EM: wong\_inbox@yahoo.com
AF: San Jose State University, Dept of Electrical Engineering, San Jose, CA 95192-0084 United States
AU: Marouf, E A
EM: emarouf@email.sjsu.edu
AF: San Jose State University, Dept of Electrical Engineering, San Jose, CA 95192-0084 United States
AB: The radio occultation technique was first used to study Saturn's rings through their effects on quasi-monochromatic radio signals transmitted from Voyager 1 during its flyby of Saturn in 1980. Almost a quarter of a century later, Cassini is planned to conduct a more extensive set of radio occultation experiments during its tour of the Saturn system. Cassini enjoys the advantage of a wide range of ring viewing geometry as well as the unique new capability of simultaneously transmitting 0.94, 3.6 and 13 cm-wavelength coherent radio signals (Ka-, X-, and S-band, respectively). Observed extinction of the direct signal and time-sequence spectra (spectrogram) of the near-forward scattered signal can be used to infer the size distribution of particles of resolved ring features (among other objectives). The inference requires solving three distinct inversion problems to recover from the measurements: i) the multiply-scattered collective diffraction lobe of a resolved ring feature, ii) the first-order scattering contribution to the collective lobe, and iii) the corresponding particle size distribution. Although various classical regularization techniques may be used for this purpose, a subjective valuation of solution smoothness usually needs to be introduced. We investigate an alternative approach based on Bayesian function learning schemes which provides a rigorous probabilistic framework to address the tradeoff between data fit residuals and prior knowledge about the character of the solution. In contrast with the regularization approach, the Bayesian approach provides estimates of confidence intervals for the most-likely solution achieved, an important advantage. The approach is particularly adaptable to some Cassini occultations of relatively unfavorable alignment between contours of constant Doppler shift in the ring plane and circular boundaries of ring features, as the approach naturally "fuses" time-sequence of spectra each containing contributions from adjacent ring features. We also use the Bayesian approach to combine in a single step inversion of (simulated) extinction and diffraction lobe observations to recover the particle size distribution over the centimeter to several meters size range without assuming an explicit model. Only the first-order scattering approximation has been considered in our investigation so far, an idealization to be removed in future work.
DE: 5759 Rings and dust
DE: 6265 Planetary rings
DE: 6275 Saturn
SC: Planetary Sciences [P]
MN: 2004 AGU Fall Meeting