TEMPLATE MATCHING TO BUILD A HIGH QUALITY EARTHQUAKE CATALOG OF THE 2011 YOUNGSTOWN, OHIO SEISMIC SEQUENCE AND TEST ITS RELATION TO A LOCAL WASTEWATER INJECTION WELL

From March to December 2011, the Ohio Department of Natural Resources (ODNR) recorded 11 earthquakes in Youngstown, OH, leading to speculation that the earthquakes were being caused by a wastewater injection well, which stopped pumping after a local seismic network showed the earthquakes were nucleating near the injection point. Unfortunately, 11 events identified by ODNR plus the local deployment data represent a limited characterization of the seismic sequence, making it difficult to confirm a causal relationship between injection and the earthquakes. This is a natural limitation of traditional seismic techniques, which required an earthquake to be M2.0 to be identified by ODNR before the local deployment. To address this limitation, we developed a multiple station template matching (waveform cross correlation) algorithm, which is able to detect events ~10x smaller than traditional techniques. Our technique utilizes regional broadband seismometers located within 200km of the earthquakes, eliminating the need for costly local seismic deployments and allowing us to investigate backwards into existing time series. We detect ~280 earthquakes in Youngstown, allowing us to test the correlation between seismicity and injection. Earthquakes started two weeks after injection began and ended 2 weeks after injection ended. Relocation of earthquake hypocenters shows the first events occurring near the injection well, propagating WSW (consistent with the focal mechanism of the largest earthquake) at a rate of 1-2 km/y. Individual families of events show delay times of 1-6 days between injection volume fluctuations and seismicity rate variations, proportional to distance from the well, likely representing diffusivity in the now saturated fault zone. A change in diffusivity after arrival of the triggering front suggests development of more fully integrated fault zone permeability though time. We envision that pore-fluid pressure increased in discontinuous permeable zones of the fault system, reduced effective normal stress and permitted fault slip to occur. Early displacement could promote fluid infiltration into adjacent regions of the fault, iteratively increasing the area of potential failure, consistent with the observation that the largest events (M>2) were uncommon until late in the sequence.