Geoacoustic parameter extraction using reverberation data from the 2000 boundary characterization experiment on the Malta plateau

This paper presents some new results from measurements of seafloor reverberation and pulse spreading using horizontal and vertical line arrays. The principal objective of this paper is to extract useful geoacoustic and bottom-scattering parameters that apply over a large ocean area. Analysis is presented on reverberation data from the 2000 Boundary Characterization Experiment performed jointly with North Atlantic Treaty Organization (NATO) Undersea Research Center (NURC), Applied Research Laboratory (ARL) of Pennsylvania State University, Defence Research and Development Canada (DRDC), and Naval Research Laboratory (NRL). Sources were SUS charges and coherent pulses. The receivers were horizontal arrays used monostatically. Data were analyzed in bands from 80 to 4000 Hz. Highlights of the reverberant returns are discussed. The experiment site is the Malta Plateau area south of Sicily, a relatively flat heavily sedimented area, but with a rocky ridge to the east. An original aspect of this paper is the design and implementation of a new automated inverse method using towed-array data to accomplish that goal. For each data set, a multiple-step simulated annealing (SA) algorithm is used together with the Generic Sonar Model (GSM). After automatically adjusting bottom loss and scattering strength, good agreement is achieved between the diffuse reverberation data and model predictions in relatively flat areas. Model/data differences are generally correlated with bottom-scattering features. Since reverberation from SUS charges typically lasts 10-40 s or more, extracted parameters apply over wide areas. Independent acoustic measurements provided a basis for a comparison with extracted values. Local bottom-loss and backscattering measurements were made by Holland in these areas. Additionally, chirp-sonar measurements were analyzed by Turgut. A comparison of geoacoustic models obtained with their methods and with this one was quite good. Comparing transmission loss (TL) predicted with Turgut's local inverse method and TL predicted with the method presented here gave answers that were usually within 3 dB of each other. Typical two-way time spreads of 0.25 s were seen at a range of 7.5 km, with normalized peak correlations of 0.5, and which were fairly consistent with predictions made using the inverse results