Premium
Improving Absolute Earthquake Location in West Texas Using Probabilistic, Proxy Ground‐Truth Station Corrections
Author(s) -
Lomax Anthony,
Savvaidis Alexandros
Publication year - 2019
Publication title -
journal of geophysical research: solid earth
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.983
H-Index - 232
eISSN - 2169-9356
pISSN - 2169-9313
DOI - 10.1029/2019jb017727
Subject(s) - induced seismicity , relocation , seismology , geodesy , geology , ground truth , proxy (statistics) , structural basin , statistics , mathematics , paleontology , machine learning , computer science , programming language
An increase in induced seismicity in the central United States since 2009 led to establishment of TexNet seismic‐monitoring in Texas. Accurate, absolute seismic‐event location is critical to TexNet, allowing quantitative evaluation of possible association of seismicity with human activity. For the Delaware Basin, western Texas, relocation using different velocity models and TexNet station subsets shows absolute location error up to 4–5 km. The preferred method to reduce absolute error, ground‐truth calibration, is not available in this area. Alternatively, we used industrial well activity as proxy, ground‐truth for developing probabilistic, proxy ground‐truth (PPGT) station corrections for relocation. Assuming well activity causes seismicity, we defined a distance–time probability associating events and well activity. We used these associations and other evidence to show some seismicity in the Delaware Basin is more likely due to hydraulic‐fracturing than saltwater disposal. We then probabilistically accumulated PPGT station corrections using event hypocenters constrained to associated fracturing‐well locations. We applied this procedure within 12 km of TexNet station PB02, optimizing the procedure through comparison of rates of causal and acausal associations. Relative to the initial locations, final PPGT relocations show smaller residuals and shifts in epicenter as much as 3 km, predominantly toward the north and northwest. PPGT residuals are similar to those from relocation with standard station corrections. The initial hypocenters showed an unreasonable deepening with distance from station PB02, whereas PPGT relocations produced an overall flattening of event depths. These results are consistent with PPGT corrections giving real improvement in absolute location accuracy.