Array observations of an oscillating seismic sequence in the Reykjanes Peninsula, SW-Iceland, in December 2021

In the evening hours of 21 December 2021, a seismic sequence started in south-central Reykjanes peninsula in SW-Iceland. Both the science community and the civil protection agency were alarmed due to the proximity of this sequence to the site of the 2021 Fagradalsfjall eruption (March – September 2021), especially as this was the most prominent sequence since the end of the eruption and it showed similar characteristic as the seismic activity that had been observed in the 3 weeks leading up to it. In addition, the December earthquake sequence was located along a NE-SW striking alignment which, together with GPS and InSAR measurements, has been interpreted as a dike intrusion, which also was the origin of the March eruption. We analyse the seismic activity using a small-aperture (D=1.7 km, d=0.5 km) urban seismic array, consisting of 5 Raspberry Shake 4D sensors (1 vertical geophone and 3 MEM accelerometric components) located in the nearby municipality of Grindavík about 10 km WSW from the former eruption site. During the first days of the seismic activity magnitudes reached up to ML 4.8 but on 30 December the activity subsided and then ceased, with only few events reaching more than ML 2, which coincides with the magnitude of completeness of the seismic array.  

We present the first insights into the spatiotemporal characteristics of the sequence provided by array processing of the most intense period of the sequence. To process the array data, we used the SeisComP module AUTOLAMBDA with both the FK and PMCC (Progressive Multi-Channel Correlation) method to obtain back azimuth and slowness pairs of incoming waves. During its first hours, the sequence showed a systematic behaviour in the back azimuthal distribution of the incoming waves. Namely, over a repeated interval of a couple of hours the back azimuthal estimates increase steadily at a rate of 5 to 12°/h after which the source of the activity appears to drop back to the initial azimuthal values, and the cycle repeats. Over the following days, these bursts of oscillating activity become less frequent with relatively calm phases between. These periods of oscillating behaviour show that the seismic activity was systematically migrating southwest to/from northeast and most likely is the signature of a pulsating magma pressure front in the dike itself. This behaviour is similar to some phases during the previous eruption when lava was actively erupting with hours of quiescence in between. These results show that the monitoring of automatic back azimuth and slowness estimates are a useful tool in revealing small-scale systematic behaviour of seismic sequences in the area in real-time.