Towards 3-D geoacoustic quantification: Application of machine learning to 2-D seabed surveys
Abstract
AbstractKnowledge of seabed geoacoustic properties are needed for a variety of scientific, commercial, and military purposes. For example, they are crucial for predicting acoustic propagation in shallow water as well as bottom-limited areas on continental slopesand in the deep ocean. In this proposal, the focus is on the seabed geoacoustic properties in the upper few tens of meters and frequencies from of order hundreds of hertz to several kilohertz. The preponderance of current geoacoustic sampling methods are 1-dimensional (1-D) and can be categorized either as point measurements, e.g., cores and in-situ probes, or a single geoacoustic profile resulting from a source and receiver spaced tens to hundreds of water depths apart. These latter estimates are very useful, but to someextent are point estimates of a non-linear average over the extent of the measurement aperture, typically kilometers, with the associated biases that come from such averaging. The long-term goal of this research is to be able to measure and infer seabed geoacoustic properties and their uncertainties in three spatial dimensions at much higher resolution in lateral and vertical extent. This will substantively advance our understanding of sediment spatial variability in continental shelf environments. The specific goal of the proposed 3-year research program is to advance from stitched high spatial resolution 1-D seabed models to machine-learning accelerated quasi 2-D geoacoustic property quantification.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 12, 2023
- Source ID
- N000142312322
Entities
People
- Charles W. Holland
Organizations
- Office of Naval Research
- Portland State University
- United States Navy