Determination of fine-scale bottom roughness using combined optical and acoustical sensing.

Abstract

Seafloor characterization work in progress at the Deep Submergence Laboratory takes advantage of a multiscale suite of optical and acoustic sensors deployed from towed and remotely operated vehicles. A long-term goal is the automated segmentation and classification of undersea terrain. Fine-scale data of small areal extent are used in the interpretation and ground truthing of coarser, larger scale data sets, which in turn are applied to other measurements of increasing scale and decreasing resolution: Camera and laser bathymetry are used for the ground truthing of high-frequency (e.g., 200–500 kHz) swath-bathymetric sidelook sonars; both optical and precision acoustical data serve for the interpretation and validation of intermediate-frequency sidelook data. Under development is an iterative technique, with the goal of improving classification results. At the first iteration, knowledge of sensor and seafloor geometry is combined with acoustic backscatter data to estimate a regional scattering function. This function is then applied to the raw amplitude data for removal of a deterministic component that depends on the bottom slope. The resultant map of acoustic albedo is segmented, a local scattering function is estimated within each segment, and the process is repeated until a measure of convergence is satisfied.