We have reprocessed seven cross-axis common depth point (CDP) profiles between the Clipperton transform and the 9¿17'N Deval (deviation in axial linearity) on the East Pacific Rise (EPR) to understand the relationship between axial magma chamber (AMC) width and seafloor morphology. Forward modeling of cross-axis CDP profiles suggests a segmented AMC in which significant variations in width occur across minor rise axis discontinuities (e.g., Devals). The modeled rise segment widths bounded by the 9¿53'N--9¿35'N Devals, the 9¿35'N--9¿17'N Devals, and south of the 9¿17'N Deval were ≤0.7 km, 1.0--1.2 km, and 4.15 km, respectively. Transition in AMC width across these discontinuities is unclear due to the sparseness of cross-axis line spacing; however, a simple association of Devals with decreased magma supply is doubtful: the minimum (250 m) and maximum (4150 m) AMC widths are found near the 9¿35'N and 9¿17'N Devals, respectively.

The reprocessing of CDP profiles included repicking stacking velocities to ensure a proper stack of the AMC reflector and its associated diffractions, imaging postcritical reflection from the base of layer 2A, finite difference time migration, ray theoretical depth migration, and travel time modeling of AMC diffraction patterns. Constraints on AMC width were derived from forward modeling techniques based on analytic raytracing. Velocity models were constructed from SeaBeam bathymetry and modified expanding spread profile (ESP) velocity-depth functions. ESP velocity models were altered to compensate for off-axis thickening of layer 2A as imaged in the CDP reflection data. Continuous two-dimensional velocity models constructed from modified ESP velocity-depth functions and SeaBeam bathymetry should account for ray bending at the seafloor/basalt interface and any lateral velocity gradients induced by a thickening layer 2A. Stacked data were time migrated using a finite difference algorithm and extrapolated to depth using ray theoretical depth migration. Reflector positions were input into our forward modeling scheme to produce a zero-offset travel time response of our migrated solution.

The travel time response was then overlain on the stacked section to ensure an adequate match, especially to diffractions generated at the AMC edge. Forward modeling of AMC diffraction patterns reveals that original AMC width estimates were too large. The undermigration of the AMC reflector resulted from the conversion of stacking to interval velocities without accounting for topographic effects on individual CMP gathers, thus resulting in improperly collapsed diffraction hyperbolae. Ship wandering relative to the AMC edge can account for variations in AMC reflector amplitude and dropout on the along-axis CDP line. The continuity of the AMC appears unbroken across several ridge axis discontinuities between the Clipperton transform and the 9¿17' Deval which suggests an AMC whose along-axis dimension exceeds 75 km. Reflectivity modeling of CMP gathers suggests that the available data are consistent with a magma chamber comprising only a thin layer of melt overlying a zone of partially solidified crystal mush. Such a thin layer of melt might inhibit along-axis mixing of magmas and thereby account for variations in magma composition along the rise crest. This melt lens model for the AMC would also produce strong diffraction patterns as imaged in the CDP data. ¿ American Geophysical Union 1993