MODELING OF THE ACOUSTIC REVERBERATION SPECIAL RESEARCH-PROGRAM DEEP-OCEAN SEA-FLOOR SCATTERING EXPERIMENTS USING A HYBRID WAVE-PROPAGATIONSIMULATION TECHNIQUE

Quantitative modeling of bottom-interacting ocean acoustic waves is co mplicated by the long propagation ranges and by the complexity of the scattering targets, We employ a two-dimensional (2-D) hybrid technique combining Gaussian beam, finite difference, and Kirchhoff integral so lutions of the wave equation to simulate ocean acoustic experiments wi thin half of a convergence zone in the SOFAR channel, The 2-D modeling approach is reasonable due to the one-dimensional (1-D) velocity dist ribution in the water column and the strong lineation of the seafloor morphology parallel to the mid-ocean ridges. Full-waveform modeling of ocean acoustic data requires that the topography and the material pro perties of the seafloor are available at scales that are several order s of magnitude smaller than typical bathymetric sampling rates. We hav e therefore investigated the effects on the ocean acoustic response of a stochastic interpolation scheme used to generate seafloor models, F or typical grazing angles of the incident wave field (approximately 5 degrees-20 degrees), we found that different stochastic realizations o f the same seafloor segment (sampled at 200 m) yield an intrinsic unce rtainty of the order of 3-8 dB in amplitude and 0.1-0.3 s in time for individual prominent events in the reverberant acoustic field, Hybrid simulations are compared to beam-formed ocean acoustic data collected during the Acoustic Reverberation Special Research Program (ARSRP) cru ises, Side lobe noise in the observed acoustic data is simulated by ad ding band-limited white noise at -30 dB relative to the maximum intens ity in the synthetic data. Numerical simulations can be limited to the response of only one of the mirror azimuth beams provided that the ex perimental geometry is suitably chosen, For the 2-D approximation to b e valid, the cross-range resolution of the observed data must be small er than the characteristic scale of seafloor lineations, and the beams of interest must be approximately perpendicular to the dominant struc tural grain. Under these premises, the arrival time and maximum intens ity of the observed backscattered acoustic events can be modeled withi n 0.3 s and 5-10 dB, respectively, The mismatch in arrival time is int erpreted to be due predominantly to intrinsic uncertainties in the sto chastic interpolation of the seafloor profiles, whereas the fact that the intensity of the backscattered events is systematically overestima ted is attributed to 3-D effects within the ''footprint'' of the beam and/or to underestimated noise levels.