Refined Principal Stress Estimates from Induced Seismicity in Southern Kansas and Oklahoma Based on Seismological Tools and Laboratory Experiments

Successful storage of large volumes of CO2 in the subsurface requires improved understanding of the state of-stress at and below reservoir depth in order to mitigate the hazards associated with storage integrity and induced seismicity. However, determining stress orientations and magnitudes at storage reservoir depths is technically challenging, both vertically and laterally away from a wellbore. State-of-the-art methods of stress field estimation require direct measurements through boreholes, but these are expensive, spatially sparse, and potentially compromised simply by virtue of drilling the borehole. Moreover, direct stress field measurements made in boreholes are insensitive to the stress field at distances larger than a few hundreds of meters from the well-bore and leave an open question as to what the stresses are at the depths that earthquakes occur. Widely used elasto-tectonic models have been developed to infer stresses over regions where no direct measurements are available, but these methods fail to quantify the uncertainty. Additionally, borehole-based measurements (using wireline density logs, dipole sonic logs, borehole imaging, hydraulic fracturing tests, and other stress indicators) can only confidently measure/infer the minimum principal stress and therefore, only certain components of the full stress tensor. While, work is underway to develop a borehole tool for full stress tensor measurement, this does not address temporal changes in the stress field due to injection. In this project we are developing methodologies to measure the in-situ principal stress in the deep subsurface through use of multiple, independent, but complementary seismic methods (i.e. interferometry, shear wave splitting, and focal mechanism inversion), laboratory verification, and development of theoretical frameworks. By leveraging existing regional and local datasets we are developing a set of diagnostic tools for determining the in-situ stress state with reduced uncertainty at and below reservoir depths (1.5-6 km). We will apply remote geophysical methods such as focal mechanism inversion and shear wave splitting measurements to estimate spatial and temporal changes in the stress orientation by analyzing waveform records of local seismic and microseismic events recorded over extended time periods during times of nearby active injection. These methods can provide broader spatial coverage by sampling a range of depths and distances from a well. Additionally, they provide in-situ measurements of the stress field that do not require drilling boreholes to reach the locations of interest. We will further address poorly understood uncertainties of these indirect measurements through quantitative analysis of the seismic datasets and integration with results from laboratory experiments on basement core samples. The existing regional and local datasets used in this project are catalogs of more than 24,000 relocated earthquakes from M~0 to M4.9 induced near active fluid disposal wells in Kansas and Oklahoma. Water disposal in the subsurface is used as a proxy for CO2 injection. Applying the virtual seismometer method (VSM) we obtain a refined model of the 3D earth structure directly around the microseismic events. We can also measure variation in focal mechanisms, allowing us to track changes in stress orientation. These focal mechanism solutions are also inverted for 4D estimates of the stress tensor. Preliminary results, in Kansas, from shear wave splitting analysis of this dataset indicate that the maximum principal stress orientation is generally aligned with the expected ENE direction.