Physics-Based Forecasting of Time, Magnitude, and Probability of Induced Earthquakes in Oklahoma
Abstract
The recent seismicity surge in the U.S. midcontinent since 2008 is attributed to deep wastewater injection. Quantifying the causal link between injection operation and the seismic response is vital for hazard assessment efforts. Among many factors, injection rate, background stressing rate, and local hydrogeology are thought to be the main factors determining the magnitude-time distribution of these events. Accounting for these factors using a poroelastic model combined with a rate-and-state friction law, the changes in the crustal stress and seismicity rate due to fluid injection can be computed and thus can be further exploited for assessment of induced earthquake hazard through a statistical framework (Zhai & Shirzaei, 2018, GRL). Here, we investigate the seismic, hydrogeologic, and injection data spanning the period of 1995 - 2017 in northern-central Oklahoma. The magnitude-time distribution of the observed M3+ earthquakes for the period of 2008 - 2017 is accurately reproducible. In response to injection rate reduction in 2016, the exceedance earthquake probability of M5.0 decreases by 22% in Western Oklahoma by the year 2017, while that in Central Oklahoma remains unaffected. The different responses are attributed to faster fluid diffusion rate and larger injection reduction in Western Oklahoma compared with the Central part. After a hypothetical injection shut-in in April 2017, the earthquake exceedance probability will approach its historical background level by 2025 at both Central and Western Oklahoma. The delay between injection shut-in and exceedance probability reduction is primarily due to the time-dependent nature of fluid diffusion. We conclude that the increased fluid pressure at pre-stressed faults is the main driver of the induced earthquakes in Oklahoma. An effective induced earthquake forecasting effort requires accounting for the physics of fluid diffusion and earthquake nucleation. Employing such physics-based models for assessing time-dependent induced earthquake hazard are useful with extensive potential for operational induced earthquake forecasting.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFM.S22A..04Z
- Keywords:
-
- 7209 Earthquake dynamics;
- SEISMOLOGYDE: 7223 Earthquake interaction;
- forecasting;
- and prediction;
- SEISMOLOGYDE: 7230 Seismicity and tectonics;
- SEISMOLOGYDE: 8168 Stresses: general;
- TECTONOPHYSICS