The seismic attenuation structure of the East Pacific Rise

Abstract

Studies of seismic propagation through oceanic crust have contributed enormously to our understanding of the generation and evolution of oceanic crust However, such work has largely been confined to the seismic velocity structure. In this thesis we present results from a study of seismic attenuation using a data set collected for three-dimensional tomographic imaging of a fast-spreading ridge. The experiment location at 9°30'N on the East Pacific Rise is the site of a strong mid-crustal seismic reflector which has been inferred to be the roof of a small axial magma chamber at about 1.6 km depth. A spectral method is used to estimate t*, a measure of the integrated attenuation along a wave path. Such a method asswnes that the dominant frequency-dependent component of propagation is intrinsic attenuation. A logarithmic parameterization is then used to invert t* measurements for Q-1 structure asswning that the velocity structure is given from earlier studies. To evaluate the method of Q tomography a full-waveform finitedifference technique which does not include attenuation is used to calculate solutions for seismic propagation through a two-dimensional velocity model. The results show a complex pattern of seismic propagation in the vicinity of the axial magma chamber. The first arrival always passes above the magma chamber. However, for paths of significant length that cross the rise axis the amplitude of this arrival is very small, and the first phase with significant amplitude is a diffraction below the magma chamber. High-amplitude Moho turning and PP arrivals may also be important secondary arrivals. Synthetic inversions show the importance of selecting time windows for power spectral estimation which are dominated by a single phase and of using wave paths which closely corresponds to that of the selected phase. A comparison of the finite difference solutions and the predictions of the a twodimensional, exact ray-tracing algorithm with record sections obtained during the tomography experiment significantly improves our understanding of seismic propagation across the East Pacific Rise. The results enable an objective choice of the position and length of the time window fort* estimation. Moreover, additional constraints are incorporated into an approximate three-dimensional ray-tracing algorithm used in the inversion so that the wave paths more closely correspond to those of the desired phase. The full data set to be inverted comprises about 3500 t* estimates and includes crustal paths which do not cross the rise axis, diffractions above and below the axial magma chamber, and Moho-turning phases. Wave paths for the Moho-turning phases cross the rise axis at a wide range of lower crustal depths. The Q-1 models resulting from two-dimensional and three-dimensional tomographic inversions show that the attenuation of seismic waves on the East Pacific Rise is dominated by two regions of low Q; one in the upper 1 km of crust, and one at depths greater than about 2 km below the rise axis. While the data do not resolve the details of vertical variations in near-surface Q-1, the results show a substantial variation in shallow attenuation within 0.05 My of the rise axis. On-axis, Q values averaged over the upper 1 km are about 100, while off-axis the average value rapidly decreases to about 30. Measurements of the seismic velocity suggest that the thickness of the surficial highporosity extrusive layer increases substantially off-axis. If such thickening is entirely responsible for the observed change in near-surface attenuation then Q within the extrusive layer must be much less than 20. Alternatively, in situ changes in porosity may also contribute to the observed increase in attenuation. Since significant tectonic activity is apparently restricted to locations well off-axis we suggest that such variations in porosity may result from hydrothermal activity. Regions of hydrothermal downwelling located off-axis will be subject to cooling and thermally-induced cracking while upwelling regions on-axis may be accompanied by rapid infilling of existing pores by hydrothermal deposits. Estimates of t* for all phases propagating below the magma chamber are markedly higher than those for other phases, resulting in Q-1 models which include a region of low Q extending from 2 to 7 km depth below the rise axis. The lowest Q values resolved are about 25-30 both immediately below the magma chamber and within the lower crust. While there is some evidence for a small decrease in attenuation with depth in the lower crust, axial Q values at depths ranging from less than 2.5 to 6 km are relatively constant, always lying below 50. Laboratory measurements at seismic frequencies suggest that Q values of 25-50 require only very small fractions of partial melt. The attenuation observations thus place constraints on the dimensions of the axial magma chamber and strongly suggest that the thickness of the region containing more than a few percent of partial melt is no more than 1 km.