Automatic seismic event locations in Germany using TieBeNN

🗂️ Description This repository contains automatic absolute locations of local seismic events in Germany and surrounding regions, computed using the automatic location system TieBeNN (TiefenBestimmung mittels Neuronaler Netze). Event date times and epicenter coordinates were taken from manual catalogs to test TieBeNN. All events have: A minimum local magnitude of 1.0 A tectonic origin The following event catalogs are included: automatic_catalog_2021-2023_tiebenn.csv – 594 local events from the BGR catalog (2021–2023) automatic_catalog_2024_BENS.csv – 55 events from the BENS catalog (2024) automatic_catalog_2024_EDSW.csv – 50 events from the EDSW catalog (2024) automatic_catalog_2024_SED.csv – 50 events from the SED catalog (2024) automatic_catalog_2024_TSN_EdB.csv – 50 events from the TSN catalog (2024), located using BGR’s 1D velocity model automatic_catalog_2024_TSN_WB2012.csv – 50 events from the TSN catalog (2024), located using the WB2012 1D velocity model (Růžek & Horálek, 2013) 🔹 Abbreviations: BGR = Bundesanstalt für Geowissenschaften und Rohstoffe BENS = Erdbebenstation Bensberg SED = Schweizerischer Erdbebendienst EDSW = Erdbebendienst Suedwest TSN = Thüringer Seismologisches Netz 📊 Catalog structure Each CSV file contains: Hypocenter coordinates and uncertainties Eight normalized features used to compute the Location Quality Score (LQS) These are the columns and their descriptions: datetime: Event origin time latitude / longitude (°): Epicenter coordinates depth (km): Hypocentral depth unc_x / unc_y / unc_z (km): location uncertainty in x, y, z directions norm.det.cov: Normalized location covariance norm.sta.den: Normalized station density norm.azgap: Normalized azimuthal gap norm.sec.azgap: Normalized secondary azimuthal gap norm.aui: Normalized azimuthal uniformity index norm.near.sta: Normalized nearest station distance norm.rms: Normalized RMS of travel time residuals norm.npicks: Normalized number of phase picks LQS: Location Quality Score This new version uses an updated version of station density, which slightly modifies the LQS. Automatic locations remain the same. 🤖 About TieBeNN TieBeNN is a Python-based wrapper that integrates open-source, machine-learning tools for: Phase picking Denoising Phase association Probabilistic hypocenter estimation via NonLinLoc 🔗 Code: https://github.com/Cthuulhaa/tiebenn 🔗 DOI: https://zenodo.org/records/15384316 📘 Docs: https://tiebenn.readthedocs.io/en/latest --- 📄 Citation A pre-print of the scientific paper describing TieBeNN is now available on EarthArXiv: Ramos et al. (2025). A quasi-real-time system for automatic local event monitoring in Germany. EarthArXiv. (https://doi.org/10.31223/X5774H)