Waveform similarity for quality control of event locations, time picking, and moment tensor solutions

Abstract

Multiplet analysis assumes that events with highly similar waveforms originate in the same source region and with the same source mechanism. We have determined how waveform similarity could be used at different stages of microseismic processing and interpretation for quality-control purposes to assess internal consistency among events from the same multiplet group. Crosscorrelation (CC) values decayed with interevent distance, and we found that doublets with large separation distances revealed location errors due to mispicks. Using histograms of differential times between highly correlated events (doublets), time-picking errors could also be detected. Waveform correlation could also detect inconsistencies in derived source mechanisms. Doublets should fall within similar regions in the Hudson source-type plots even for events with large nondouble-couple components or when the inversion results were internally inconsistent. A similar assessment was done by plotting the P- and T-axes of multiplets in a focal sphere. We used a minimum CC threshold of 90%, a sufficiently high level, such that the assumption of waveform similarity remained valid, and we analyzed a microseismic data set recorded during two stages of a hydraulic fracturing experiment to demonstrate these quality-control procedures. These techniques could be used in other settings, such as geothermal studies, mining projects, reservoir monitoring, and earthquake seismology and were most beneficial in scenarios in which repeating events were expected, facilitating the detection of compounded errors commonly encountered in microseismic analysis.