Calibrated locations for induced earthquakes in the Delaware Basin, Texas, spanning 2009-2017

The spatiotemporal association between seismicity and oil and gas production in the Delaware Basin (DB) has been publicly well-monitored since the Texas Seismological Network (TexNet) went operational in 2017. Prior to TexNet, the regional network geometry in the basin changed but indicated an increase in the seismicity rate since 2009. These studies, including TXAR dataset (Frohlich et al., 2020) used here, report epicentral uncertainties for pre-2017 events on the order of >15 km. Equally poor depth control for pre-2017 catalogs complicates the association of seismicity with hydraulic fracturing and both shallow and deep saltwater disposal (SWD) in the DB. We provide updated relative locations for pre-2017 seismicity using a multiple event relocation algorithm (MLOC), which ties the original earthquakes in space to 39 well-located post-2017 events, known as the core cluster. The procedure ensures the hypocenters are minimally biased by unknown structure and effectively removes outliers through calibration. MLOC iterates in each run, starting with calculating the hypocentroid (the reference point) using the core cluster dataset (readings at < 80 km distance) and relocating each event with respect to the hypocentroid. Relative errors in the core cluster are limited to 1 km and depths are constrained by close local readings. The hypocentriod errors, as a measure of absolute errors for the whole cluster, are 0.4 and 0.5 km in latitude and longitude, respectively. Stepwise addition of pre-2017 events to the core cluster is followed by calibration based on empirical reading errors. Depth for events in 2009 are fixed from close readings at Transportable Array stations but is a free parameter in the inversion for the rest of the pre-2017 catalog. We introduce S-P differential arrivals for single events with parent phases that have been flagged out. We also use cross-correlation derived differential travel times for the pre-2017 catalog. Our preliminary results show a good spatial correlation between post- and pre-2017 DB earthquakes in Reeves county, Texas. Relocated pre-2017 events have relative epicentral uncertainties of 2-5 km. Depth estimates from the inversion systematically are shallower than the initial TXAR catalog with a smaller spread around 2 km that is more consistent with shallow (1-4 km depth) SWD in the DB.