Both volcanism and faulting contribute to the rugged topography that is created at the Mid-Atlantic Ridge (MAR) and preserved off-axis in Atlantic abyssal hill terrain. Distinguishing volcanic from fault-generated topography is essential to understanding the variations in these processes and how these variations are affected by the three-dimensional pattern of mantle upwelling, ridge segmentation, and offsets. Here we describe a new quantitative method for identifying faut-generated topography in swath bathymetry data by measuring topographic curvature. The curvature method can distinguish large normal faults from volcanic features, whereas slope methods cannot because both faults and volcanic constructs can produce steep slopes. The combination of curvature and slope information allows inward and outward facing fault faces to be mapped. We apply the method to Sea Beam data collected along the MAR between 28¿ and 29¿30'N. The fault styles mapped in this way are strongly correlated with their location within the ridge segmentation framework: long, linear, small-throw faults occur toward segment centers, while shorter, larger-throw, curved faults occur toward ends; these variations reflect those of active faults within the axial valley. We investigate two different physical mechanisms that could affect fault interactions and thus underlie variations in abyssal hill topography at the MAR. In the first model only one fault is active at a time on each side of the rift valley.

Each fault grows while migrating away from the volcanic center due to dike injection; extension across the fault causes a flexural rotation of nearby inactive faults. The amount of stress necessary to displace the fault increases as the fault grows. When reaching a critical size the fault stops growing as fault activity jumps inward as a new fault starts its growth near the rift valley. This model yields a realistic terracelike morphology from the rift valley floor into the rift mountains; the relief is caused by the net rotation accumulated in the lithosphere from the active faults (e.g., 10¿ reached 20 km from the active fault). Fault spacing is controlled by lithospheric thickness, fault angle, and the ratio of amagmatic to magmatic extension. We hypothesize that this mechanism may be dominant toward ridge segment offsets. An alternative model considers multiple active faults; each fault relieves stresses as it grows and inhibits the growth of nearby faults, causing a characteristic fault spacing. Such fault interactions would occur in a region of necking instability involving deformation over an extended area. This mode of extension would drive a feedback mechanism that would act to regulate the size of nearby faults. We hypothesize that this mechanism may be active in the relatively weak regions of strong mantle upwelling near segment midpoints, causing the homogeneous abyssal hill fabric in these regions. ¿ American Geophysical Union 1993