CORRELATION AND THE LACK THEREOF: INJECTION, EARTHQUAKE, AND MAGNITUDE DISTRIBUTIONS IN OKLAHOMA

While it is important and appropriate for us, collectively, to look for positive correlations in seismicity patterns to evaluate seismic hazard, it is equally important to recognize, acknowledge, and advertise when observations do not support established models. Regionally and temporally, the large increase of injection volumes into the Arbuckle Group strongly correlates with the increase in earthquake occurrence. Locally, the correlation does not hold up. The highest injection volumes do not correlate well with either the greatest earthquake densities or the largest earthquakes. Further, the spatial distribution of relatively large earthquakes (MW ≥ 4.5) does not correlate well with the spatial distribution of smaller earthquakes. Many of the highest magnitude-weighted spatial densities of all earthquakes in the period 2011-2016 (eastern Oklahoma County, eastern Woodward County, southern Logan County, eastern Logan County, and southern Grant County) have not experienced a MW ≥ 4.5 event. Most of the MW ≥ 4.5 events have occurred in areas with relatively lower magnitude-weighted spatial earthquake densities. That is: Relatively high magnitude-weighted spatial earthquake densities do not correlate with larger events. Further, neither larger earthquakes nor high magnitude-weighted earthquakes densities correlate well with long-term injection volumes. Analyzing the purported precursors to larger events and sequences (2016 Pawnee, 2016 Galena Township) provides no meaningful, useful tools to forecast the likelihood, size, or location of future earthquakes. We must acknowledge the correlation of increased earthquake rates with the increase of produced-fluid disposal into the Arbuckle Group, and the apparent effectiveness of reducing earthquake rates by reducing injection rates (whether by market forces or regulatory action). But we must also acknowledge that the observations challenge our fundamental models.