Correlation between Changes in Seismicity Rates and Well Injection Volumes in Oklahoma

Abstract:

We present a statistical approach to establish correlations between locations with seismicity increase in Oklahoma and nearby well injection volumes. Seismicity rates in the state have significantly increased since approximately 2008. Fluid injection into deep wells has been theorized to be the cause of this seismicity, but the increase occurred significantly after the start of injection activities in the region. Further, injection-induced earthquakes depend on the presence and orientation of basement faults and the stress state in the region. Because of these complexities, it has been difficult to directly correlate fluid injection with seismicity. Here we show that a statistical correlation between increase in seismicity and injection volumes can be established in Oklahoma. We first employ a change point method to locate the regions where a change in seismicity rates has occurred. We then use a logistic regression model to relate the injection volumes in a region with the presence or absence of seismicity change in the region. This model is further used to evaluate the relative contribution of cumulative injection volumes and monthly injection rates to seismicity. The model can be used to identify “seismically sensitive regions” where seismicity increase has been observed with little fluid injection, and “seismically stable regions” where seismicity changes have not been observed even with high fluid injection. This information can be combined with geological information in a region, and used to make decisions about acceptable volumes for injection and to identify lower-risk regions for injection.