The thickness and internal properties of the magma sill located at the top of the axial magma chamber (AMC) along the southern East Pacific Rise (EPR) have been investigated through a combination of waveform modeling the near-vertical incidence reflections from this body and analysis of reflection amplitude variation as a function of source-receiver offset (or slowness). Our results show that the AMC reflector observed along the southern EPR is best modeled by a thin (< 100 rn thick) sill of partial melt (Vs • 0 km/s) sandwiched between higher-velocity material, and that the thickest sills are generally associated with the lowest P and S wave velocities. The comparatively high P wave velocities and nonzero shear wave velocities inferred for this sill indicate that it is filled with partially molten magma which in some locations has a high crystal content. This may have important implications for eruption mechanisms and along-axis mixing of magma at the EPR. There is no simple relationship between morphologic indicators of magma supply (e.g., axial depth or volume) and sill thickness, depth, or velocity. Magma sill properties may be closely tied to the eruption and replenishment cycle of the AMC and thus may vary on a much shorter spatial and temporal scale than axial morphology, which reflects longer-term variations in magma supply to the ridge. These studies have led to a model in which a thin (< 200 m), narrow (typically < 1 km wide) lens or sill of magma 1-2 km below the seafloor overlies a zone of partial melt in the midcrust that is in turn surrounded by a broader low-velocity volume (5-10 km wide) extending to the base of the crust [Sinton and Derrick, 1992]. Although the magma sill at the roof of this low-velocity body may be volumetrically quite small, it is thought to play a key role in the availability and composition of magma at the rise axis [Sinton and Derrick, 1992]. A t Also at MIT/WHOI Joint Program in Oceanography, Woods Hole, Massachusetts. 2Now at Paper number 96JB01907. 0148-0227/96/96JB-01907509.00 clearer understanding of its properties and their variation along axis will provide additional insight into the genesis of oceanic crust. waveform relative to the seafloor reflection [Barth, 1987; Derrick et al., 1987; Vera et al., 1990] are consistent with the large negative impedence contrast expected between magma and overlying crustal rocks. Migration and forward modeling of diffraction hyperbolae in unmigrated cross-axis reflection profiles, generated by the abrupt termination of the sill, indicate a typical sill width of 500-1000 m [Kent et al., 1990, 1993, 1994]. Estimates of the thickness of this body are not well constrained. Tomographic velocity models [Toomey et al., 1990, 1994] and seismic attenuation studies [Wilcock et al., 1992] place an upper limit of 1-2 km on the thickness of any largely molten, midcrustal body beneath the EPR. The identification of a second, deeper reflection in waveform studies of the AMC indicates the presence of a thin magma sill with a thickness of only a few...

[1] The upper crustal seismic structure of the slow spreading Mid-Atlantic Ridge is studied using a genetic algorithm-based waveform inversion of multichannel streamer data. Four single-ship multichannel profiles from 35°N are analyzed: one in the rift valley and three in the rift mountains along 0.7, 1.6, and 1.9 Ma crust. A layer 2A horizon is continuously imaged along three profiles and is associated with a sharp velocity increase from extrusives to dikes. Its depth and regularity in the rift valley indicate that most of the extrusive section is built on the inner valley floor through a pattern of deposition and fault-bounded uplift into the rift mountains. Its variability along one line, however, shows that this process is disrupted during tectonically dominated periods. A thickening of layer 2A toward the Oceanographer fracture zone may be the result of along-axis magma transport. The interval of rapid velocity increase at the base of layer 2A thins with age, a possible response to enhanced hydrothermal mineralization within the zone of mixed dikes and extrusives. Transition zone ($200 m) and off-axis layer 2A thicknesses (350-600 m) are similar to those at other spreading centers. This indicates that equivalent extrusive volumes are produced at all spreading rates along a relatively narrow zone of dike emplacement. However, differences in on-axis layer 2A thickness between this area and fast spreading ridges suggest that the exact pattern of thickening varies between spreading regimes. Relative to fast spreading ridges, the moderate velocity increase with age recorded in the upper crust (from 2.3 to >2.7 km s À1 within $2 Myr) may be due to a less active hydrothermal system and hence slower porosity reduction. INDEX TERMS: 3035 Marine

Abstract. Magnetic field measurements made by submersible define the cross-sectional geometry of a magnetic polarity reversal boundary and the vertical variation of crustal magnetization in upper oceanic crust. Measured polarity boundaries show a systematic pattern of shallow dip towards the spreading axis within the upper extrusive lavas, and steeper dip in the lower extrusive lavas. This geometry is a consequence of the emplacement of extrusive lava at a midocean ridge. Reversal boundary geometry and magnetization estimates are used to calculate the magnetic contribution of the extrusive lava sequence to the overlying marine magnetic anomaly signal. From the forward modeling, the highly magnetized extrusive lavas contribute the majority (50-75%) of the observed sea surface magnetic anomaly, consistent with the extrusive crust forming the primary source layer for young marine magnetic anomalies.

[1] The detailed upper crustal structure of the East Pacific Rise (EPR) at 17°20 0 S is examined by applying a genetic algorithm-based waveform inversion to five multichannel seismic lines: one on-axis and two to either side along 42-and 85-kyr-old crust. On-axis, a double-stepped velocity pattern is recorded beneath 70-100 m of lowvelocity extrusives (2.1-2.4 km s À1 ). We define the upper velocity contrast as the base of seismic layer 2A due to its severity and continuity along and across axis. The more subdued and intermittent lower-velocity step is not observed off-axis. Material between the two high-gradient intervals is proposed to represent the pillow/dike transition, bounded above by a sharp increase in dike fraction with depth and below by an abrupt change in rheology and/or deformation. Extrusive velocities increase quite rapidly in this area, with velocities $3 km s À1 common in crust 85 kyr old. This, plus a rapid (300-400 m) thickening of layer 2A observed within 1-4 km of the rise axis, indicates that this segment is undergoing focused melt delivery (<500-m-wide dike intrusion zone) and elevated hydrothermal activity. These findings demonstrate the ability of single-ship multichannel data to record detailed information on the reflectivity and velocity of the upper crust and the ability of the genetic algorithm to efficiently construct accurate seismic models based on this information. INDEX TERMS: 3035 Marine