Shallow-water matched-field localization off Panama City, Florida

Abstract

The results of a shallow-water localization experiment, performed 19 miles south of Panama City, Florida in October 1985, are presented. The experiment involved a 450-Hz source placed 2.2 km from a vertical array of 16 hydrophones in ∼33 m of water. The experimental site was essentially range independent with a flat, hard, sandy bottom. Successful passive localization of the source was obtained using a maximum-likelihood matched-field processor. Studies were undertaken to determine the robustness of the localization to variation of the following parameters: water depth, sediment sound-speed profile, sediment density and attenuation, and array tilt. It was found that, in order to ensure localization accuracy and robustness, the environmental parameters important to know well are the water depth, sediment sound speed, and array tilt. However, the matched-field processor is much more tolerant to inaccuracies in estimates of the sediment density and attenuation. This corresponds clearly with the results of two simulation studies by DelBalzo et al. [J. Acoust. Soc. Am. 83, 2180–2185 (1988)] and Feuillade et al. [J. Acoust. Soc. Am. 85, 2354–2364 (1989)]. A further conclusion drawn from the experiment is that localization in a shallow-water waveguide using maximum-likelihood processing is complicated by the repetitive sidelobe structure of the acoustic field. This suggests that time-domain and other broadband localization techniques may achieve better localization performance because of their additional frequency-averaging capability.

Inversion for geometric and geoacoustic parameters in shallow water: Experimental results

“…In all three cases the sidelobe level was very close to that of the source peak. In 1990, Feuillade et al [5] reported a single shallow water localization in very shallow water (33 m) using the Capon maximum likehood processor. In 1993, Jesus [6] reported on successful source localizations for transient signals.…”

“…The so-called 'matched-field' method simply implies that acoustic field observations are 'matched' in some sense to multiple iterations of a forward model as a function of search parameters. This form of processing has been applied to a variety of estimation problems including source localization, see for example [1][2][3][4][5][6][7], tomography [8], and inversion for ocean and ocean bottom properties [9][10][11][12][13]. In many cases the 'matching' is accomplished via a correlation process often referred to as the linear or Bartlett processor.…”

Matched Field Processing in Ocean Acoustics

“…Years later, Feuillade et at. [33] reported the results of an MFP experiment with full field replicas in the same Gulf of Mexico region where they localized a source at a range of 2.2 km from a 16 element vertical array in 33 m of water. Both conventional and MVDF processors were used, the latter displaying the great ambiguity of a repetitive sidelobe structure.…”

Marine information gathering, transmission, processing, and \\fusion: current status and future trends

“…20 世纪 80 年代以来, 在声信号处理中引入传播模型的一种备受关注的技术是匹配场处理, 其最 初主要用于解决声源定位问题 [48,49] . 然而, 匹配场处理对模型误差非常敏感, 存在所谓的 "失配" 问 题 [50,51] , 包括环境、系统和统计失配 3 个方面. 大量研究旨在解决该问题, 例如 Richardson 等在匹 配场处理算法中加入先验信息, 得到声源位置的后验概率密度函数, 用来表示未知声传播参数的不确 定度 [52] .…”