A technique suggested by Hales et al. (1981) for determining hypocentral depths that are relatively independent of the velocity model originally used to locate the earthquakes is applied to a carefully selected set of earthquakes located with the Adak Seismic Network. The origin times are taken from the Wadati diagram for each event and the epicentral distances from the standard hypocenter solution are assumed to be correct. A straight-ray interpretation, based on the use of a single average velocity from the hypocenter to the surface, is then used to determine the average velocity and the depth from the slope and intercept of the linear X2 − T2 plot.
From the Adak catalog for August 1974 to May 1983, 8046 located events, 579 earthquakes with good P-data and 347 with good S-data were selected on the basis of excellent time picks, good quality solutions, and reliable Wadati and X2 − T2 plots. The revised depths show that the sharp cutoff of the shallower main thrust zone activity at 26 km seen in the Adak data set is an artifact of the routine location procedure. Routine depths are found to be systematically too shallow by a mean amount of 4.35 km, with origin times 0.20 sec too early.
The velocities derived in this study agree with those in the standard model for depths below 80 km, but are higher than the standard values at shallower depths. The two models can be reconciled by invoking small gradients of velocity in the uppermost three layers, but this result is not unique. The results indicate local variations of the shallower velocities within the Adak seismic zone, one of which, low velocities beneath the volcanic arc, has been found in an earlier, independent study.