A recent SeaMARC II survey along the flanks and crest of the East Pacific Rise between 13¿ and 15 ¿N (Edwards et al., 1988) provides us with a unique opportunity to compare stochastic models of the seafloor against bathymetric data. Stochastic models can be used to quantitatively characterize intermediate to small-scale (<100 km) seafloor features such as abyssal hills. In a recent paper, Goff and Jordan (1988) introduced a model for the two-point covariance function having five free parameters which describe the amplitude, orientation, characteristic wavelengths, and Hausdorff (fractal) dimension of seafloor topography. These parameters can be estimated from multibeam data such as Sea Beam by inverting auto- and cross-covariance functions for the beams.

The stochastic model can be used to generate synthetic topography to arbitrary resolution and scale. By assuming appropriate Sea Beam noise and response characteristics we can use the synthetic topography to produce synthetic data sets. Comparison of real and synthetic data is an important means of assessing the performance of the stochastic model. Unfortunately, the limited spatial extent of Sea Beam in the across-track direction and lack of 100% coverage over large areas severely limit our ability to compare the full two-dimensional stochastic model with real seafloor morphology, especially with respect to abyssal-hill length characteristics. The recent SeaMARC II survey included sufficient off-axis topography (over 100 km, 70,000 km2) to permit a comparison of a complete two-dimensional synthetic topographic field with a region of abyssal hill terrain that has close to 100% data coverage. We have estimated stochastic models for various regions of the SeaMARC II survey area from several Sea Beam tracks which cross it. Synthetic fields are generated from these models using a Fourier method. These comparisons indicate the extent to which the model succeeds in characterizing seafloor topography, and the directions we will need to take to improve our modeling. A change in ridge complexity at 14 ¿N and the location of a propagator pseudofault appear to be correlated with changes in stochastic character modeled in this region. Such correlations may provide important new constraints on the understanding of ridge processes. ¿ American Geophysical Union 1991