This project seeks to demonstrate a capability to conduct wide area mapping of the distribution of munitions at shallow underwater sites (≤ 20 m depth), using commercial-off-the-shelf (COTS) highfrequency sonar technology and advanced signal processing techniques. The goal is to quantify and improve the probability of detection of unexploded ordnance (UXO) that may be found on the bottom while providing the geomorphological context necessary to estimate the extent of site contamination based on predictions of munitions mobility and burial. The long-term goal is to produce distribution maps of munitions for site managers to inform remediation decisions.
Researchers have developed an in situ beam pattern estimation technique that achieves range and cross-range normalization while preserving the acoustic backscatter intensity measurement. Researchers will calibrate the backscatter measurements using standard reference targets (e.g., tungsten carbide sphere). The resulting acoustic backscatter intensity image will be used for target identification as well as sediment characterization, the latter being key to providing the necessary context for munitions distribution. Automated target recognition (ATR) in seafloor acoustic imagery relies on sophisticated image processing techniques to identify targets of interest from the shape of their echoes and shadows. The first step is an image normalization that seeks to remove variability in the local mean background level from the image. While such techniques are effective in highlighting targets in the scene, they eliminate contextual information because image pixel values are no longer related to the underlying acoustic backscattering process. This project will demonstrate the in situ beam pattern estimation technique that preserves the acoustic backscatter intensity using survey data obtained with a COTS high-frequency (200-400 kHz) dual-head multibeam sonar system. A demonstration survey area along the Mississippi coast in the northern Gulf of Mexico will be seeded with known targets and surveyed with the sonar system. In addition, continuous environmental data will be collected during the survey. The output will be simultaneous geo-referenced and co-registered spatial grids of bathymetry and fully corrected bottom acoustic backscatter intensity with a horizontal spatial resolution on the order of 1% of the altitude of the sonar. Researchers will then adapt existing ATR and texture mapping algorithms to delineate areas of the survey where munitions were (1) detected on the bottom, (2) not detected and unlikely to be present, or (3) not detected but potentially present subbottom.
This effort will provide a new method that combines bathymetry and acoustic backscatter imagery to produce high-fidelity maps of concentrations of munitions while simultaneously identifying regions that are free of munitions over wide areas in shallow water. The key advantage of producing corrected acoustic imagery containing the absolute backscatter intensity is that the observable differences in the imagery are no longer relative with an arbitrary scaling. Consequently, a properly corrected acoustic image contains much more contextual information than imagery currently being used for detection and classification. This project will benefit the DoD by reducing the number of false alarms per survey, with associated savings in fewer subsequent target prosecutions. The demonstration will initially focus on swath mapping sonars operating at 200-400 kHz that are well-suited to surveying in shallow water ranging from 10 m to 20 m depth. For water depths less than 10 m, the method is unambiguously scalable to higher frequency sonars (> 1 MHz), likely best deployed from unmanned or autonomous vehicles. Researchers expect that the technique is also adaptable to swath mapping sonars operating at about 30 kHz, where some penetration in the subbottom is known to occur.