Seafloor parameterization using simple reflection quantities in matched field inversion

摘要

In practical matched field geoacoustic inversion, the parameterization for the seabed is the critical first step. However, to overcome the the complexity of the actual sea bottom and lack of the priori information, the popular solution at present is to form an effect-equivalent description of the bottom using several geoacoustic parameters to characterize the acoustic properties of the seafloor. Furthermore, in the premise of containing sufficient information, inversion complexity decreases with the number of parameters. In some cases, the rapid extraction of geoacoustic information about the seafloor by a simple seabed parameterization technique is more pragmatic. In this paper, two quantities were proposed, referred to as the slope of bottom loss versus grazing angle FdB and the equivalent velocity ce. This is based on the observation that, at low grazing angles, the reflection loss and phase exhibits an approximately linear relationship to the grazing angles of incidence in a wide range of seafloor types, while long range propagation is dominated by seafloor interactions at small grazing angles. Earlier similar reflective parameter models and the relevant theory were briefly reviewed. The bottom reflection properties were expressed by just two reflection parameters, whose advantage and limitation were also discussed. To test the performance of different seafloor models when there were mismatches between the reality and model, a complex multi-layer simulation bottom was set up. Numerical simulations of matched field processing based on genetic algorithm were performed. Both reflection quantities and traditional geoacoustic parameters were inverted. Through an importance sampling procedure based on a semi-empirical approach, their posteriori probability distributions (PPD) were also estimated. The parameter sensitivity was examined by analysis of the Bartlett processor output and PPD, and the model mismatch was also discussed. The transmission loss predicted from two reflective quantities which were derived as inversion products, agreed well with the actual situation. That confirmed the validity of this parameterization technique. The results of simulation analysis showed that seafloor parameterization using simple reflection quantities allows matched field inversion to infer the geoacoustic information with a relatively easier procedure.