A statistical discussion of Pn residuals from explosions

Abstract

The problem of the statistical treatment of large numbers of Pn residuals is considered, using travel times to stations in California and Nevada from 43 explosions in Nevada. Techniques of data selection and weighting of observations are developed in order to estimate the distribution of residuals and to estimate station adjustments to Jeffreys-Bullen travel times. A new simple method of estimating the mean of a normal population from a truncated sample is given. An approximate analysis of variance showed a significant variation from one explosion to another in the mean Pn travel time. On the average, large magnitude explosions, detonated in metamorphic rock produced shortest, and most reliable, travel times. The station adjustments also differed significantly, ranging from −0.0 seconds at Fresno to −1.3 seconds at Calistoga; confidence intervals about the station adjustments had half lengths of the order of half a second. The results have implications for seismic studies which depend upon estimates of such station adjustments.