Evaluating the effects of velocity models and array configuration on induced seismic event locations in the Permian Basin

Abstract

In the final months of 2022, two substantial earthquakes with magnitudes exceeding ML 5.0 hit West Texas, USA, causing widespread public concerns about the effect of industrial activities on human and environmental safety. The monitoring of earthquakes in this region is largely based on the data of the public regional seismic array, which consists of stations spaced typically tens of kilometers apart. Accurate hypocenter determination of these induced events is crucial as it provides insight into the triggering mechanism, enabling operators and regulators to develop effective mitigation strategies. However, there are debates regarding the accuracy of the publicly reported induced event data due to discrepancies between public reports of event hypocenters and those determined using local dense arrays operated by private companies. The primary objective of this study is to identify the underlying causes of the discrepancies between results obtained from the public regional array in West Texas and local dense arrays. Through modeling and analysis of field data collected by a local dense array in the Permian Basin, we determine that three critical factors influence the reliability of induced event results. First, the accuracy of the velocity model used for the event location is the most crucial factor. Second, the distance between a station and an event plays a crucial role in determining the sensitivity of the data to the hypocenter depth. Finally, consistency between the observed and modeled wave propagation behavior is crucial for ensuring the validity of the objective function in the inversion. Our findings indicate that it is challenging to obtain reliable hypocenters using a regional sparse array with station spacing on the order of tens of kilometers. The best practice for obtaining an accurate event hypocenter and magnitude is to monitor induced seismicity using a local dense array and process the data with a velocity model that fully characterizes the local basin’s geology.