Item

Abstract

Earthquakes are commonly located by linearized inversion of discrete arrival time picks made from signals recorded at a network of seismic stations. If mis-picks are made, these will contribute to the location, therefore causing potential bias. For data recorded by a dense seismic array, direct imaging methods can be applied instead. We describe the ‘coalescence microseismic mapping’ method, which is a bridge between the two approaches and will operate with seismic data recorded continuously on a sparse array. By continuously mapping scalar signals derived from the envelope of seismic arrivals we derive robust estimates of the spatiotemporal coordinates of the origins of seismic events. Noisy data are migrated away from the correct origin, so do not contribute to errors in location. The method is rooted in a Bayesian formulation of event location traveltime inversion, allows imaging of source locations and has the capacity to handle errors in modelled traveltimes. It has the advantage of working with any 3-D velocity model, which thereforemay include anisotropy. It also automatically incorporates both P- and S-wave data. A multiresolution grid search leads to an efficient implementation, with a search over a larger domain including joint inversion for location and velocity structure possible where warranted by the data quality. We discuss the theory and implementation of this method and illustrate it with real data from microseismic events in Iceland caused by melt intrusion in the crust.