Earthquakes Induced by Wastewater Injection, Part II: Statistical Evaluation of Causal Factors and Seismicity Rate Forecasting

Abstract

Wastewater disposal has been reported as the main cause of the recent surge in seismicity rates in several parts of central United States, including Oklahoma. In this article, we employ the semi-empirical model of the companion article (Grigoratos, Rathje, et al., 2020) first to test the statistical significance of this prevailing hypothesis and then to forecast seismicity rates in Oklahoma given future injection scenarios. We also analyze the observed magnitude–frequency distributions, arguing that the reported elevated values of the Gutenberg–Richter b-value are an artifact of the finiteness of the pore-pressure perturbation zones and a more appropriate value would be close to 1.0. The results show that the vast majority (76%) of the seismically active blocks in Oklahoma can be associated with wastewater disposal at a 95% confidence level. These blocks experienced 84% of the felt seismicity in Oklahoma after 2006, including the four largest earthquakes. In terms of forecasting power, the model is able to predict the evolution of the seismicity burst starting in 2014, both in terms of timing and magnitude, even when only using seismicity data through 2011 to calibrate the model. Under the current disposal rates, the seismicity is expected to reach the pre-2009 levels after 2025, whereas the probability of a potentially damaging Mw≥5.5 event between 2018 and 2026 remains substantial at around 45%.

Time-dependent seismic hazard and risk due to wastewater injection in Oklahoma

“…Seismicity rate forecast (in red) for model calibration through the end of 2017 by Grigoratos et al (2020b) for the Area of Interest (AOI p ) shown in Figure 2, assuming that the disposal rates (in blue) remain constant after December 2017. The observed seismicity rates from the declustered catalog through mid-2019 are also shown in black.…”

“…Grigoratos et al (2020b) tested the statistical significance of this prevailing hypothesis, namely that wastewater disposal is the driving force behind the seismicity rate changes, and found that 85% of the seismicity above M w 3% and 91% of the seismicity above M w 4 can be indeed associated with wastewater disposal at a 95% confidence level. These earthquakes occurred within a large area which they defined as an Area of Interest for wastewater disposal (AOI p ; Figure 2).…”

“…Therefore, it is clear that the earthquake occurrence in Oklahoma in the last decade is a highly spatially variable time-dependent process directly linked to oil and gas activities that cannot be modeled using traditional Poissonian recurrence models. To address this problem, we previously developed a seismicity rate model that incorporates the monthly wastewater disposal rates as the main driver for the changes in seismicity rates (Grigoratos et al, 2020a). This semi-empirical model was inspired by different statistical and hydromechanical modeling approaches found in the literature, while introducing novel considerations in various steps of the process.…”

“…Non-declustered seismicity and cumulative, spatially distributed wastewater disposal volumes in December 2017 for injection between January 2006 and December 2017 within the investigation zone. The black polygon indicates the Area of Interest (AOI p ) for wastewater injection, as defined by Grigoratos et al (2020b).…”

“…In addition, we compare our loss estimates with the paid insurance claims and perform sensitivity analyses to assess our confidence in the results. Our integrated risk model employs our time-dependent earthquake rate model (Grigoratos, 2020; Grigoratos et al, 2020a), a region-specific ground motion model (GMM; Zalachoris and Rathje, 2019), a recent Vs30 map (Zalachoris et al, 2017), HAZUS exposure data (Federal Emergency Management Agency (FEMA), 2013, 2018), and updated vulnerability curves for both structural and nonstructural elements (Silva et al, 2020). The calculations are performed using a stochastic Monte Carlo–based approach (Barbat et al, 1996; Cramer et al, 1996; Crowley and Bommer, 2006) implemented in the OpenQuake engine (Pagani et al, 2014; Silva et al, 2014), an open-source software package developed by the Global Earthquake Model (GEM; http://www.globalquakemodel.org).…”