#Dataset Initial earthquake catalogs for this study were provided by a group at GNS which is tasked with collecting, and preliminary analysis of, continuous seismic data from the MRP seismograph networks. The MRP data is collected roughly every three months directly from the data loggers. These data are supplemented by data from the aforementioned GeoNet stations. GeoNet data is sampled at 100 Hz while the MRP network sampled at 200 Hz. For most of this study, these data were resampled at 50 Hz to lower computational costs.
The initial GNS earthquake catalog included more than 2000 earthquakes which were automatically detected and located by the SeisComP3 software package. As automatic S-phase picking was not implemented, these events were located with P picks only. This condition holds for the entirety of this study. We filtered the catalog to include only events within a 50 km by 50 km bounding box centered on the geothermal fields and filtered again for events with an average pick residual within 1 standard deviation of the mean event-averaged pick residual across the dataset. The locations of the resulting 637 earthquakes in the context of the seismic network are shown in Figure \ref{Figure1}. Note that the ratio of Rotokawa to Ngatamariki seismicity still holds at 3:1 for the filtered catalog of template events shown.
Mighty River Power provided well locations and pumping rates for injection wells within the Rotokawa and Ngatamariki geothermal fields for the entire study period.