ESTIMATING GEOACOUSTIC BOTTOM PROPERTIES FROM TOWED ARRAY DATA

Estimating the seabottom geophysical structure from the analysis of ac oustic returns of an explosive source (air-gun, sparker,...) has been used for a longtime as a routine survey technique. Recent work showed the possibility of using well-suited numerical models to invert the ac oustic held for estimating detailed geoacoustic sediment properties. C ommon implementations used long synthetic aperture arrays (up to 2 km and more) in order to resolve potential environmental ambiguities of t he acoustic field. Others, used vertical arrays of sensors covering a significant part of the water column to identify the channel normal mo de structure and thus gather information for the bottom physical relev ant properties. This paper investigates, with simulated data, the conc ept of using a moderate aperture physical line array and a sound sourc e simultaneously towed by a single ship for inverting the bottom geoac oustic structure from the acoustic returns received on the array. Firs t, bottom parameter estimators are derived and their system sensitivit y is investigated. In particular, it is shown that such a system may b e used to sense compressional and shear velocities on the bottom first layers. Density and attenuations (both compressional and shear) have in general small influence on the acoustic field structure and are the refore difficult to estimate. Increasing the signal frequency bandwidt h by incoherent module averaging has no significant influence on sensi tivity Mismatch cases, mainly those related to array/source relative p osition, showed that deviations of more than lambda/3 in range and lam bda/5 in depth may give erroneous extremum location and therefore bias ed final estimates. Second, two bottom parameter estimators are compar ed and their performance tested on a typical shallow water environment . In order to solve the underlying multiparameter inverse problem, glo bal search optimization is used. In particular, it is shown that the u se of an adaptive genetic algorithm may, in conjunction with a well-su ited maximum likelihood based parameter estimator, rapidly converge to the surface extremum. Inversion results are in agreement with the pre dictions obtained from the sensitivity study. The mean relative error at 10 dB signal-to-noise ratio is within 1% for the compressional velo city, while greater errors are reported for the shear velocity. Compar ison with recent results obtained with a radial basis functions (RBF) inversion strategy showed similar performance. Finally, results obtain ed with a 156 m aperture towed array showed a good agreement between t he inverted compressional velocities and the ground truth measurements .