S U M M A R Y Anisotropic inversions of surface wave data show that the variations in vertical shear velocity, pv, and anisotropy of the oceanic upper mantle in the Pacific are much smoother and more systematic functions of the age of the seafloor than has been reported in previous studies. The data used in this analysis are the pure-path results of previous studies on the lateral distribution of fundamental-mode Love and Rayleigh wave phase velocities (Nishimura & Forsyth 1985. The pure-path models include parameters which describe the variations with age, azimuthal anisotropy and residual depth anomalies. The calculated velocity models of the upper mantle are constrained to vary smoothly with depth and to represent the minimum deviation from an isotropic starting model. Inversions were performed using the method of Tarantola & Valette (1982).The two best resolved parameters of the computed transversely isotropic model are the shear wave velocity terms, fiv and 5. The results indicate that pv above 200 km progressively increases as a function of the age of the seafloor with the pattern qualitatively mimicking isotherms of theoretical thermal cooling models. If one selects the depth to the maximum negative gradient in shear velocity as being the best available indicator of lithospheric thickness, then the thickness increases from about 15-35 km beneath 0-4 Myr old seafloor to 70-110km in the oldest seafloor. The magnitude of the shear wave anisotropy term, 5, rapidly increases in the first 20 Myr until some apparent constant value is reached in the older regions. A more realistic upper mantle structure is calculated using a priori information on the correlation between changes in shear and compressional wave velocities and the expected nature of the anisotropy. The general results are the same as the previous inversion without a priori constraints. Finally, the effect of attenuation is included, the primary result being an overall increase in 6". The maximum change occurs at around 150 km depth, which reduces the velocity contrast between the lithosphere and asthenosphere. It is therefore more difficult to make a distinction between the plate and low-velocity zone when the effect of attenuation is included. An estimate of the azimuthal anisotropic structure is obtained by inverting for the Rayleigh wave cos 2v coefficients using derivatives calculated by the method of Montagner & Nataf (1986). The reference frame used to constrain the azimuthal effect is that of fossil seafloor spreading direction. The results indicate that in regions of the Pacific less than 80 Myr in age, there is significant anisotropy down to 200 km depth. In regions older than 80 Myr, azimuthal anisotropy is confined to the upper SO km. The transverse and azimuthal anisotropy structures can be explained by an oceanic upper mantle containing olivine with different orientations.

The lateral distribution of fundamental-mode Rayleigh wave phase velocities in the Pacific has been calculated in order to determine the variation in velocity as a function of the age of the oceanic plate, the importance of azimuthal anisotropy, and the presence of secondary lateral heterogeneities. The data set used in this study consists of phase velocity measurements in the period range 20-125 s for 178 paths traversing the Pacific. Both the pure-path and spherical harmonic inversion techniques are used in this investigation with an emphasis placed on determining the effectiveness and resolving power of these methods in calculating the velocity distribution in an oceanic regime.The first-order velocity-age relation which dominantly characterizes surface wave dispersion in an oceanic environment can be adequately modelled by either the pure-path or spherical harmonic techniques. Azimuthal anisotropy is the dominant second-order effect and is best developed in regions less than 80 Myr in age. In this age zone, the faster direction of wave propagation is either in the direction of fossil seafloor spreading or in the direction of present-day absolute plate motion. In the western Pacific, there is a correlation with paleo-relative motion (fossil seafloor spreading) for periods less than 50 s. Lateral velocity heterogeneities which are not related to the age of the oceanic plate are found by pure-path inversions and are correlated with regions associated with shallow residual depth anomalies. This effect (slow velocities with respect to the velocity-age model) monotonically decreases as a function of period and is indicative of a shallow origin. Age independent anomalies were also modelled by the application of the pure-path and spherical harmonic methods in a sequential inversion.

We present a revised model of the tectonic evolution of the Juan de Fuca ridge by propagating rifting. The new model has three different relative rotation poles, covering the time intervals 17.0–8.5 Ma, 8.5–5.0 Ma, and 5.0 Ma to the present. The rotation pole shifts at 8.5 and 5.0 Ma imply clockwise shifts in the direction of relative motion of 10° to 15°. At each of these shifts, the pattern of propagation reorganizes, and the new ridges formed by propagation are at an orientation closer to orthogonal to the new direction of motion than the orientation of the preexisting ridges. The model, containing a total of seven propagation sequences, shows excellent agreement with the isochrons inferred from the magnetic anomaly data, except in areas complicated by the separate Explorer and Gorda plates. The agreement between model and data near the Explorer plate breaks down abruptly at an age of about 5 Ma, indicating that the probable cause of the rotation pole shift at that time was the separation of the Explorer plate from the Juan de Fuca plate.