Digging deeper with template matching: challenges and opportunities in seismicity analysis

Earthquake catalogs have traditionally been constructed using P- and S-picks determined with short-term and long-term signal averages. Template matching consists of using the detected catalog events as templates to detect similar but smaller events. It has been used in recent years with great success to identify typically an order of magnitude more earthquakes than conventional techniques. Catalogs produced with this technique have the potential to transform future analyses of seismicity due to the exceptional temporal and spatial resolution in both foreshock-mainshock and aftershock sequences as well as in swarms. This resolution has enabled us to differentiate between the role of afterslip and pore fluid diffusion in the 2010 Mw7.2 El Mayor aftershock sequence. Earthquake source studies of stress drops as well as repeating events are also greatly enhanced using these new catalogs.

These catalogs also come with a new set of challenges that have yet to be explored in detail, in part because we have added a much larger set of events that are of ultimately lower quality than the original templates. These challenges may require new developments in magnitude determination, earthquake locations, as well as statistical seismology. Here, we discuss some of the challenges and limitations of this technique, which involve magnitude completeness and inherent biases, spatiotemporal variation of detection rates, magnitude estimation in weakly recorded events, and double-difference relocation in datasets of heterogeneous quality. We will show how some of these problems can be mitigated if considered beforehand, such as explicitly treating template events differently from detections for relocation purposes. These subjects will be discussed in the context of a recently completed template matching catalog for southern California (2008-2017) with more than 2.4 million earthquakes.