Velocity model determination for accurate location of mining-induced seismic events

Abstract

Determining the location of mining-induced seismic events is strongly dependent on having an accurate velocity model. However, such a model is seldom available. This paper describes the determination of a velocity model for seismic event location, using the seismic events themselves as sources whose location is to be determined along with the parameters of the velocity model (a simultaneous inversion of event locations and velocity structure). Seismic monitoring of a mine in Colorado is used as an example, with an array of geophones installed both on the surface and in underground roadways. Velocity models of increasing complexity are considered, starting with a homogeneous velocity, moving to a (slightly dipping) layered-earth model, and eventually including static time shifts to account for the effects of a weathered, near-surface, low-velocity layer on arrival times at geophones mounted on the surface, This series of increasingly complex models obviously shows increasingly better fits to the data, but also shows more plausible event locations, and with more realistic elevation spans. Examination of spatial patters in the residuals indicates that there are likely mining-induced changes in velocity that are not accounted for in the model.