Review of Local and Regional Seismicity in the Carpathian Basin Using Multiple Event Location Algorithms with ground truth events

Over the years, we have made significant updates to the Hungarian National Seismological Bulletin as velocity models, localization techniques, and station network configurations have evolved. A crucial aspect of our work involves addressing anthropogenic events since many recorded events initially classified as earthquakes are actually mining explosions, complicating geological interpretations. Since the Kövesligethy Radó Seismological Observatory has been collecting digital data since 1995, this is an ideal time for us to review the entire catalog using contemporary algorithms and velocity models.

By 2018, we had recalculated hypocenter parameters for 5,735 events (Bondár et al., 2018) using the 3D RSTT velocity model, which produced reliable results for Hungary. We have since expanded the dataset with an additional 6,578 events through December 2022, ensuring the inclusion of the highest-quality initial hypocenter parameters.

Our research has two main components: a comprehensive analysis of seismicity across the Pannonian Basin using the Bayesloc algorithm (1995-2022) and the precise relocation of specific event clusters using the double-difference method (e.g., Somogyszob, Szarvas, Móri-árok). Before conducting the Bayesloc analysis, we reviewed event types and identified hundreds as potentially anthropogenic. We also performed quality control, filtering events to retain those with favorable station geometry for accurate initial estimates.

For the Bayesloc analysis, we used GT2 events as reference points. Before finalizing results, we tested and documented the impacts of initial hypocenters, GT2 events, and a priori standard deviations on hypocenter parameters. This analysis included refining travel-time corrections, phase identification, and precision metrics to improve accuracy.

Our results showed reduced location errors and clear clustering, particularly for GT2 events, enabling more reliable geological interpretations. For local event clusters, the double-difference algorithm proved highly effective for small-scale studies, using differential times from waveform cross-correlation to achieve optimal relative positioning.

In this project, we made over 25 years of data compatible for simultaneous analysis, yielding the most reliable results for the Pannonian Basin to date and enabling improved seismic hazard assessments. Identifying and excluding anthropogenic events is crucial for accurate geological interpretation and seismic risk assessment. Our workflow also supports annual database updates and consistent processing of local event clusters.