In this work, we cross‐correlated waveforms in a global dataset consisting of over 310 million waveforms from nearly 3.8 million events recorded between 1970 and 2013 for two purposes: to better understand the nature of global seismicity and to evaluate correlation as a technique for automated event processing. We found that about 14.5% of the events for which we have at least one waveform correlated with at least one other event at the 0.6 or higher level. Within the geographic regions where our waveform holdings are complete or nearly complete, that fraction rose to nearly 18%. Moreover, among the events for which we had one or more seismograms recorded at distances less than 12°, the fraction of correlated events was much higher, often exceeding 50%.
These results imply that global seismicity contains a large number of repeating events, that is, events that are sufficiently similar to each other to have correlated waveforms over the time period spanned by our dataset. These results are very encouraging for using correlation in aspects of automated event processing. It is well known that because of the strongly implied similarity of the sources of correlated signals, they can be used as empirical signal detectors (ESD) to detect, locate, and identify an event using as few as one channel. Our results are very encouraging for using correlation and perhaps other forms of ESD for regional network processing and continental global processing because, for example, nearly all continental seismicity (99%) is within 12° of at least one International Monitoring System station.