Spatial variability of seabed properties on the Malta Plateau inferred with an autonomous underwater vehicle

We develop an automated method to infer geoacoustic properties along tracks surveyed by autonomous underwater vehicles (AUVs). The AUV tows a 32-element receiver array and a source emitting signals ("pings") at regular intervals. Recordings are processed in terms of reflection coefficients, resulting in large data volumes with substantively more information on seabed structure than traditional seismic profiling. However, interpreting seabed spatial variability requires efficient inversion. The inverse problem is non-linear and requires Bayesian sampling to quantify parameter uncertainties. To account for changes in the number of seabed layers at each ping position, the parametrization treats this number as unknown with a Poisson prior and even-numbered order statistics to improve efficiency. The method is applied to 340 data sets along a 14-km track on the Malta Plateau, employing 8 graphics processing units for approximately 2 weeks of computing time. The results resolve layering along the track with previously unreported detail. An erosional boundary is clearly resolved as a high-velocity, high-density layer and appears rougher and is buried deeper in shallower water. Depressions along this boundary are filled in with lower-velocity material. In addition, sound attenuation is well constrained in a thick low-velocity wedge. [Work supported by ONR and SERDP.]

from #ENT via xlomafota13 on Inoreader http://ift.tt/1TrRfXC
via IFTTT