Joint relative location of earthquakes without a pre-defined velocity model: an example from a peculiar seismic cluster on Katla volcano's south-flank (Iceland)

Relative locationmethods are commonly used to precisely locate earthquake clusters consisting of similar waveforms. Repeating waveforms are often recorded at volcanoes, where, however, the crust structure is expected to contain strong heterogeneities and therefore the 1-D velocity model assumption that is made in most location strategies is not likely to describe reality. A peculiar cluster of repeating low-frequency seismic events was recorded on the south flank of Katla volcano (Iceland) from 2011. As the hypocentres are located at the rim of the glacier, the seismicity may be due to volcanic or glacial processes. Information on the size and shape of the cluster may help constraining the source process. The extreme similarity of waveforms points to a very small spatial distribution of hypocentres. In order to extract meaningful information about size and shape of the cluster, we minimize uncertainty by optimizing the cross-correlation measurements and relative-location process.With a synthetic test we determine the best parameters for differential-time measurements and estimate their uncertainties, specifically for each waveform. We design a location strategy to work without a pre-defined velocity model, by formulating and inverting the problem to seek changes in both location and slowness, thus accounting for azimuth, take-off angles and velocity deviations from a 1-D model. We solve the inversion explicitly in order to propagate data errors through the calculation. With this approach we are able to resolve a source volume few tens of metres wide in horizontal directions and around 100 metres in depth. There is no suggestion that the hypocentres lie on a single fault plane and the depth distribution indicates that their source is unlikely to be related to glacial processes as the ice thickness is not expected to exceed few tens of metres in the source area. Our method is designed for a very small source region, allowing us to assume a constant slowness for the whole cluster and to include the effects of 3-D heterogeneity such as refraction. Similar circumstances may arise in other volcanic regions with a high level of heterogeneity and where densely clustered earthquakes are often recorded.