Statistical Characterization of Bathymetry and Stratigraphy on Continental Margins

A "realistic" interpolation or extrapolation of bathymetry, i.e., one that honors both the statistical character and the deterministic constraints of the data, is referred to as a "conditional simulation." The first step in generating a conditional simulation is derivation of a statistical model for bathymetry. We typically employ the anisotropic von Kármán model (Goff and Jordan, 1988), which has proven appropriate for a wide variety of geophysical fields. However, refinements of this model have occasionally proven necessary in the shelf setting, including the shelf northern California shelf morphology (Goff et al., in press). In such cases, a superposition of two models (e.g., a Gaussian covariance at larger scales, and a von Kármán model at smaller scales) is usually satisfactory. Once an appropriate statistical model has been selected, a conditional simulation can is generated typically either Fourier simulation or sequential Gaussian simulation.