Nowcasting Earthquakes: A Comparison of Induced Earthquakes in Oklahoma and at the Geysers, California

Abstract

Nowcasting is a new method of statistically classifying seismicity and seismic risk (Rundle et al. 2016). In this paper, the method is applied to the induced seismicity at the Geysers geothermal region in California and the induced seismicity due to fluid injection in Oklahoma. Nowcasting utilizes the catalogs of seismicity in these regions. Two earthquake magnitudes are selected, one large say $M_{\lambda } \ge 4$, and one small say $M_{\sigma } \ge 2$. The method utilizes the number of small earthquakes that occurs between pairs of large earthquakes. The cumulative probability distribution of these values is obtained. The earthquake potential score (EPS) is defined by the number of small earthquakes that has occurred since the last large earthquake, the point where this number falls on the cumulative probability distribution of interevent counts defines the EPS. A major advantage of nowcasting is that it utilizes “natural time”, earthquake counts, between events rather than clock time. Thus, it is not necessary to decluster aftershocks and the results are applicable if the level of induced seismicity varies in time. The application of natural time to the accumulation of the seismic hazard depends on the applicability of Gutenberg–Richter (GR) scaling. The increasing number of small earthquakes that occur after a large earthquake can be scaled to give the risk of a large earthquake occurring. To illustrate our approach, we utilize the number of $M_{\sigma } \ge 2.75$ earthquakes in Oklahoma to nowcast the number of $M_{\lambda } \ge 4.0$ earthquakes in Oklahoma. The applicability of the scaling is illustrated during the rapid build-up of injection-induced seismicity between 2012 and 2016, and the subsequent reduction in seismicity associated with a reduction in fluid injections. The same method is applied to the geothermal-induced seismicity at the Geysers, California, for comparison.