Earthquake location and pick information from 1979 to 2017 at all depths were downloaded from the Northern California Earthquake Data Center (NCEDC) in a box surrounding Mammoth Mountain (Fig. 4, boundaries marked by a blue dotted line). Earthquakes were excluded if they had both a depth above 6 km and a magnitude of <M1. Pick information and waveforms for earthquakes with catalog locations deeper than 6 km below sea level from 1995 to 2017 were used to generate templates for finding additional detections in continuous data with a matched-filter approach (16) using the Python open-source package EQcorrscan (18). Waveforms and pick information from a temporary deployment of broadband seismometers around the summit during 2012–2014 (14) were used where available. We markedly reduced computational overhead by only searching for matches to a given template 24 hours before and after its occurrence; most of the seismicity at these depths occur in swarms or bursts lasting several hours, and we were primarily interested in improving detections during these swarms rather than fully documenting their occurrence (or apparent lack thereof) between swarms. We separately ran a subset of templates through a year of continuous data and found that ~70% of unique matches occurred during the bounds of the swarm the templates belonged to and closer to 98% of the highest quality matches. For this work, we used 1394 template events with 3 to 34 channels of data and used a detection threshold of 10*(median absolute deviation)—a measure of when a detection is well above the noise—in 1-hour chunks. Further, we required the average normalized cross-correlation coefficient for a detection to exceed 0.4, which was mostly enforced during periods where the median absolute deviation was artificially lowered due to missing data. Given previous observations of reversed waveforms at Mammoth Mountain (15), we explicitly allowed for negative matches. Under these criteria, we generated 6346 additional detections.

Temporary stations are not shown. The dotted blue box shows the extent of the original NCEDC earthquake catalog search, and the red box corresponds to the map view shown in Fig. 1. The inset map also shows two distant stations used in focal mechanism determination.