Shallow water tomography in a highly variable scenario

C. Soares, S.M. Jesus , 
SiPLAB-FCT, Universidade do Algarve
8005-139 Faro, Portugal
E. Coelho
NATO Undersea Research Centre
Via San Bartolomeu, 400
19138 La Spezia, Italia

Comments: download pdf file
in  Acoustics Inversion Methods and Experiments for Assessment of the Shallow Water Environment, (Springer) Caiti, Chapman, Hermand and Jesus (eds.),  Ischia, Italy, June 2004.

: Ocean acoustic tomography was initially proposed for deep water regions where the ray approximation was valid and sound speed could be analytically linked to acoustic ray travel-time. Travel-time based tomography turned out to be highly dependent on the ability to separate closely spaced arrivals and on the precise knowledge on the source-receiver relative position at all times. Instead, matched-field tomography (MFT) is based on some sort of correlation of the full pressure field to the signal received at an array of sensors and only requires relative travel times to which an approximate knowledge of the source-receiver position is sufficient. In most operational shallow water scenarios, only MFT is applicable due to the close arrivals from bottom and surface reflections and to a perpetual source/sensor motion. In October 2000, SiPLAB and the Instituto Hidrografico (IH - PN) conducted the INTIFANTE'00 sea trial in an area off the Peninsula of Troia, approximately 50 km south from Lisbon, in Portugal. The experiment itself and results obtained in part of the data set have been reported during these last two years in various occasions. This paper focuses on the data acquired during Event 2, where acoustic propagation path was approximately range independent and the source ship was held on station at constant range of 5.5 km from the vertical line array (VLA). Although these conditions look like ideal for MFT, retrieval of water column and bottom parameters over an 14 hours long recording reveal itself extremely difficult. This paper analyses in detail the characteristics of that data set and in particular proposes a selection of the data snapshot, prior to inversion, according to the variable source range (due to the source ship drift), according to tidal point where the data was acquired and to the phase correlation at each frequency. It is shown that this data selection procedure provides a better understanding of the influence of the environment and thus final inversion results that are highly consistent with independent measurements.

ACKNOWLEDGMENT: this work was partially supported by ATOMS, FCT - Portugal.