We constrain the subduction geometry of the Aleutian slab by the distribution of seismicity recorded at local seismic networks and by a teleseismic P wave residual-sphere analysis of 11 earthquakes and one nuclear explosion. The source depth and mechanism of each earthquake are determined by modeling teleseismically recorded long-period P and SH waveforms. We construct theoretical residual spheres by tracing rays through three-dimensional thermal models of the subducting slab assuming a pressure-independent thermal coefficient of velocity (partial-V(p)partial-T). Rays are initiated at the estimated source depths and at epicenters whose relative locations are consistent with those obtained from the observed travel times after correction for slab effects. For each event, we determine the values of slab strike and dip that provide the best match to the observations. Average values of 67% confidence intervals determined by an F test are +/- 5-degrees in dip and +/- 10-degrees in strike. Strike varies along the arc by 60-degrees. For all but one of the modeled events, the estimated orientation of the slab's aseismic extension agrees to within two standard deviations of that derived from seismicity distributions. Assuming a constant dip, a slab penetration depth of 600 km fits the central Aleutian events best. This is at least 300 km below the deepest earthquakes. Slabs terminating above 400 km or below 1000 km provide significantly worse fits. Though there is a trade-off between partial-V(p)partial-T and penetration depth, the best-fitting value of partial-V(p)partial-T, using a 600-km-deep slab, is -0.5 +/- 0.1 ms-10K-1. This is consistent with both ultrasonic laboratory measurements and values obtained for deeper earthquakes in several other subduction zones. Partial-V(p)partial-T estimated from shallow thrust events is 20-50% larger than that for intermediate-focus events suggesting that a simple thermal model with constant partial-V(p)partial-T under predicts the velocity anomalies shallower than 100 km depth. The fact that a 600-km-deep slab provides better fits than a 1000-km-deep slab implies that the downdip P rays exit the slab near a depth 600 km. A slab bend to either a shallower or steeper dip would satisfy the P times. Hand-picked PcP travel times are most consistent with the slab bending to a 20-degrees steeper dip before plunging into the lower mantle. In the western Aleutians, the relative plate motion becomes transform, seismicity ceases at less than 100 km depth and vollcanism is no longer active. We image a slab-like, high-seismic velocity zone apparently associated with the along-arc lateral transport of cold subducted lithosphere from the east. An earthquake whose location is known from local observations is analyzed in detail. The velocity structure inferred using this location is nearly identical to that obtained using our scheme of annihilating location information from both theoretical and observed times. A severe bias would result, however, from the commonly used procedure of analyzing travel times computed with respect to an incorrect event location, such as that of the ISC.
Ex668Times Cited:24Cited References Count:52