Seismic travel times for central Asian epicentres

Abstract

Observations of P and S from deep earthquakes in the Hindu Kush region are analysed. They are reduced to a surface focus by a method of S. Mohorovičić. Cubic forms in Δ are found to fit the times of P satisfactorily up to 19° and of S to 21°. The coefficient of the cube term is less for S than for P, indicating that the S rays are much straighter than the P ones. A cubic from 21° to 37° and a quadratic from 37° to 95° fitted the times of P, with t and dt/dΔ continuous. Beyond 92° it seems that the time curve becomes nearly straight, as had been suspected previously. For S cubics from 22.5° to 37° and from 37° to 94° fitted; there are few data beyond 95° and a linear extrapolation was made to 105°. For both P and S there are considerable drops in dt/dΔ about distance 20°. Analysis for P at 1° intervals showed that the times beyond 20° vary smoothly with distance, and that the 20° discontinuity cannot be smoothed over more than 2°. The large amplitudes in this range strongly suggest that the variation of dt/dΔ is continuous and not due to the entry of an overtaking branch. The data for S are less clear but would suggest a similar strong d2t/dΔ2 about 22°. There is evidence that the times of P, extrapolated to distance 0°, are about 0.6 s more than those found by Willmore in the Heligoland explosion. The intervals between P and its near reflexion agree with this, and could be interpreted as an effect of the height of the region. If so it may be desirable at some stage to adapt the results to sea level.