Moment Tensor inversion with Full-Waveform Inversion and Distributed Acoustic Sensing on a subglacial volcano: Grímsvötn, Iceland.

We present the results and current challenges of an experiment with Distributed Acoustic Sensing (DAS) on Grímsvötn volcano in Iceland, and its potential for improvements in source characterization using Full-Waveform Inversion (FWI). We deployed a 12 km long fibre-optic cable for one month (May 2021) on Grímsvötn, Iceland’s most active volcano on a centennial time scale, which is covered by the Vatnajökull ice cap. The cable was trenched 50 cm into the ice, following the caldera rim and ending near the central point of the caldera on top of a subglacial lake. We discover previously undetected levels of microseismicity, and we locate these events with first-arrival times and a probabilistic inversion using the Hamiltonian Monte Carlo algorithm. The ~2000 detected events have local magnitudes between -3.4 and 1.7, and their locations indicate clear clusters of activity. These appear near surface expressions of the caldera fault, such as fumaroles, and cauldrons. We use the results, combined with an initial velocity model, to set up an FWI workflow, that takes the complex topography and subsurface into account. We create the initial velocity model for the firn-ice transition by inverting a dispersion curve of the snow cat driving along the cable, and we combine this with homogeneous media for the bedrock, ice layer, and subglacial lake. We perform an initial forward modelling study for a selection of ~600 events in the frequency range of 1.5 – 3 Hz, combined with a least-squares inversion to obtain the moment tensor components of each event. Finally, we assess the resulting moment tensors and their resolution matrix in order to evaluate the potential to invert for source characteristics. This research shows the results and potential of DAS on active subglacial volcanoes, and we hope that this research will contribute to the fields of volcano monitoring and DAS modelling.