D. Mauuary
Inst. Fur Meereskunde, Dusternbrooker Weg 20, D-24143 Kiel, Germany
The identification problem in multipath ocean acoustic tomography is not only one of the more crucial but also one of the more difficult signal processing problems to solve for further inverse studies. It is only recently that a tool, based on the Bayesian decision theory and close to the data association problem in RADAR, has been proposed by Mauuary and Moura. It fundamentally uses prior information ocean variability which transits through the ray acoustic model. It also statistically solves the ray identification problem with a Bayesian strategy. Despite the inherent complexity of resulting algorithms, a first successful attempt has been made on a French tomographic set of data (GASTOM). Some simplifications and further experimental use are now being investigated on the Mediteranean set of data (THETIS). With another approach given by Send, those are the only practical tools, but, they are sufficiently general to solve the identification problem in the most difficult conditions given by unresolved and unstable data. Both solutions are very close to the generalized Kalman filter theory and joint use of these algorithms in data assimilation models can be expected.