Abstract:

Krafla is one of the five central volcanoes of the Northern Volcanic Zone in north-east Iceland and has been utilised for decades for geothermal energy production. Thus, the volcano and its geothermal system have been monitored and imaged extensively with various geophysical methods to better understand this complex geological setting not only for scientific, but also for industrial interests.

With a ten-year dataset of 30.000 manually picked seismic events from a local permanent 12 station seismic network owned by Landsvirkjun and operated by Iceland GeoSurvey, and a very dense temporary array of 98 seismic nodes deployed for one month in 2022 in the center of Krafla caldera, we imaged P- and S-wave velocity structures of the volcano by using local earthquake tomography and analysed the relocated seismicity patterns.

The velocity structures retrieved in the high-resolution 3D models for P- and S-wave velocities offer a glimpse into the subsurface of the volcanic system with the two wave types being responsive to distinct rock/fluid properties and their phases. The relocated seismicity underscores active structures pinpointed through the tomography. The seismogenic zone hosting the largest, rather diffuse cluster of earthquakes at Krafla is located at the interface of high to low Vp/Vs close to where magma was repeatedly encountered by wells. Even though these events are located at the same boundary, their focal mechanisms vary widely from double-couple mechanisms with normal and thrusting earthquakes striking in different directions, to non-double-couple explosions and implosions. To decipher if events can be attributed to different sources, we use an unsupervised machine learning approach to cluster the events based only on the waveform of the P-onset, to make sure that effects related to different propagation paths in the clustering are minimized. With this approach, events originating from diffuse seismicity clouds can be attributed to different sources, using existing focal mechanisms, available GPS data and variations in the re-injection rates at wells of the geothermal powerplant.

By applying this method to the ten-year data set, we hope to gain a better understanding of when and where structures are active, and thus offer insights if volcanic forcing such as inflation/deflation or external forcing such as regional seismicity and anthropogenic influence trigger certain seismicity patterns.

All are welcome.