Hypocenter-velocity structure coupling in local earthquake tomography

Abstract Local earthquake tomography involves the use of seismic-wave arrival times from earthquake sources with uncertain locations for the inference of three-dimensional variations in seismic velocity. Formally, this inverse problem contains explicit coupling between the earthquake hypocenters and the velocity structure. However, many studies fail to treat the hypocenter-velocity structure coupling explicitly. An examination of previous studies provides evidence that this failure can lead to significant bias in the resulting velocity model. A set of simple hypothetical test cases clearly demonstrates the seriousness of the potential bias when coupling is neglected, producing models that underestimate the actual velocity perturbations by about a factor of four. Under the same conditions, strict simultaneous inversion and simultaneous inversion with parameter separation both produce accurate inverted models. The use of appropriate station corrections and iteration of the solution can improve the results when the coupling is ignored, but velocity anomalies will still be underestimated by about a factor of two at best.