### abstract

- A stochastic model for the emplacement of dikes and lava flows at a fast spreading ridge can generate an upper oceanic crustal structure similar to that observed in seismic data from the East Pacific Rise (EPR), in ocean drill holes, and in ophiolites. In this model the location of successive dike intrusion events relative to the ridge axis is determined by a Gaussian probability function and the cumulative flow lengths of the erupted lavas are chosen to build a Gaussian-shaped lava pile, We interpret wide-angle seismic reflections from the steep velocity gradient at the base of seismic layer 2A to be the extrusive/sheeted dike contact. Seismic data from the northern and southern EPR place constraints on the on-axis extrusive layer thickness (230 +/- 50 m), the distance over which the thickening of the extrusive layer occurs (width of the accretion zone = 1-3 km) and its off-axis thickness (300-800 m) Ophiolites and ocean drill holes (DSDP Hole 504B) provide additional estimates of the thickness of the extrusive layer and constrain the thickness of the transition region from extrusives to sheeted dikes (similar to 100-200 m). A simple stochastic emplacement model, where the lavas are described by one mean flow length, fits the thickening of the extrusive layer off-axis inferred from the deepening of seismic layer 2A, but the predicted transition from sheeted dikes to extrusives is too thick In order to match the dimensions and flat-topped shape of the seismic layer 2A boundary as well as the thickness of the extrusive-sheeted dike transition, a bimodal distribution of lava flows is used. Short flows, confined within the axial summit caldera (ASC), build up approximately half the extrusive volume. Occasional voluminous flows spill out of the ASC, or erupt outside of the ASC, and pond at a considerable distance off-axis to build up the remainder of the extrusive section. The upper part of the final extrusive section will be dominated by the off-axis flows, while the lower portions will be primarily composed of short flows erupted within the ASC. Magnetic transition widths predicted from the overlap of lavas (similar to 2 km) in this model are similar to those measured in deep-tow studies. Assuming a smoothing function which acts over one seismic wavelength, the upper crustal velocity structure predicted by the bimodal lava emplacement model is consistent with the shallow seismic velocity structure measured on the EPR, The ages of seafloor lavas in this model are younger than the tectonic spreading model ages by similar to 30-70 kyr, in agreement with anomalously young lava ages obtained from radioisotope dating of seafloor basalts near the EPR.