Abstract:
The ratio of individual P and S travel times from observatory data is considered as an indicator of velocity anomalies premonitory of large earthquakes. While this method involves model dependency in the derivation of the origin time, the origin time appears in the top and bottom of the ratio, which reduces the effects of errors.
After considering properties of the Wadati plot and travel time ratio, the values for Vp/Vs given by the ratio method, denoted, from New Zealand data are studied. These are shown to be consistent in themselves, and to concord well with values for the velocity ratio obtained by other means.
The statistical distribution for in general is considered. The distribution has a strong central tendency, and is nearly symmetric, with a few high valued outliers. While large data sets fail tests for normality, for many situations the normal distribution will be a useful approximation.
Finally, identification of anomalies is discussed. The geometric probability situation is considered for a one dimensional path passing through three dimensional anomalies, and a probability distribution derived for the length of a random line intersecting a spherical anomaly. This indicates that the bulk of intersections will be associated with large earthquakes Using synthetic data the relative travel times method is shown to have advantages over other approaches. A technique used to identify changes in mean levels in quality control situations is applied to New Zealand data to locate anomalous changes in. There is insufficient data available yet to make decisive conclusions, but this approach appears to have potential for locating anomalous changes associated with large earthquakes without prior information on when or where the change occurs.
