Relocating Earthquakes using Seismic Networks on Small Islands: Case Studies from Heimaey (Iceland) and Jan Mayen (Norway)
MetadataShow full item record
The main objective of this thesis is to study different aspects of solving the earthquake location problem and to investigate difficulties in locating earthquakes with sparse networks. This was done by using different velocity models, various ray tracing algorithms, and different location rocedures at sparse oceanic island stations. The aim is to investigate how all these factors affect the solution to the earthquake location problem on the two islands Heimaey (Iceland) and Jan Mayen (Norway). Finding the answer to this research question required that work was carried out through different steps. First, a temporary seismic network was installed on Heimaey. After this, noise analysis and a detection algorithm were applied to the recorded data. Because too few events were recorded on Heimaey, a new data-set from Jan Mayen was included in the research. The data was preprocessed, i.e., arrival times of P- and S-phases were selected in SEISAN. Subsequently, ray tracing algorithms were written for the different velocity models in the two regions. This was done by solving the ray equations numerically using the modified Euler method for the smooth models and using Snell's law for the discontinuous layered velocity model. Two different grid search methods were implemented to solve the earthquake location problem. The first method is called the single-difference grid search method. A traditional grid search method was implemented for comparison. The traditional grid search method uses calculated travel times and origin times to estimate the earthquake hypocenter, whereas the single-difference only requires travel time differences. The relocations obtained from these methods are compared to the catalog locations found by the HYPOCENTER location program available in SEISAN. In addition, the grid search methods can be used in uncertainty analyses. This is illustrated for a few events. Finally, a new velocity model for Jan Mayen, which was obtained from a 2D velocity profile in the literature, was tested. The arrival time picks of the Jan Mayen data-set were used to relocate 259 events observed after the magnitude 6.7 earthquake on November 9, 2018, and two before. The main result from the relocations was that most events were concentrated in six swarms localized along the fracture zone. These swarms were not as clear after applying the standard HYPOCENTER algorithm in SEISAN. The grid search methods localizes the events more to the Jan Mayen Transform Fault north of Jan Mayen. The location and timing of one of the swarms north-west of Jan Mayen suggests that the main event triggered it. The seismic activity along the fault returns to regular background seismicity after approximately one month. The ray tracing algorithm for the new 3D velocity model was used to relocate one of the swarms using the two grid search methods. These relocations estimated the swarm further away from the fracture zone, some kilometers south-west compared to the catalog locations. This deviation was probably caused by inaccuracies in the velocity model.
PublisherThe University of Bergen
Copyright the Author. All rights reserved