TY - JOUR A1 - Nooshiri, Nima A1 - Saul, Joachim A1 - Heimann, Sebastian A1 - Tilmann, Frederik A1 - Dahm, Torsten T1 - Revision of earthquake hypocentre locations in global bulletin data sets using source-specific station terms JF - Geophysical journal international N2 - Global earthquake locations are often associated with very large systematic travel-time residuals even for clear arrivals, especially for regional and near-regional stations in subduction zones because of their strongly heterogeneous velocity structure. Travel-time corrections can drastically reduce travel-time residuals at regional stations and, in consequence, improve the relative location accuracy. We have extended the shrinking-box source-specific station terms technique to regional and teleseismic distances and adopted the algorithm for probabilistic, nonlinear, global-search location. We evaluated the potential of the method to compute precise relative hypocentre locations on a global scale. The method has been applied to two specific test regions using existing P- and pP-phase picks. The first data set consists of 3103 events along the Chilean margin and the second one comprises 1680 earthquakes in the Tonga-Fiji subduction zone. Pick data were obtained from the GEOFON earthquake bulletin, produced using data from all available, global station networks. A set of timing corrections varying as a function of source position was calculated for each seismic station. In this way, we could correct the systematic errors introduced into the locations by the inaccuracies in the assumed velocity structure without explicitly solving for a velocity model. Residual statistics show that the median absolute deviation of the travel-time residuals is reduced by 40-60 per cent at regional distances, where the velocity anomalies are strong. Moreover, the spread of the travel-time residuals decreased by similar to 20 per cent at teleseismic distances (>28 degrees). Furthermore, strong variations in initial residuals as a function of recording distance are smoothed out in the final residuals. The relocated catalogues exhibit less scattered locations in depth and sharper images of the seismicity associated with the subducting slabs. Comparison with a high-resolution local catalogue reveals that our relocation process significantly improves the hypocentre locations compared to standard locations. KW - Seismicity and tectonics KW - Computational seismology KW - Subduction zone processes KW - Pacific Ocean KW - South America Y1 - 2016 U6 - https://doi.org/10.1093/gji/ggw405 SN - 0956-540X SN - 1365-246X VL - 208 IS - 2 SP - 589 EP - 602 PB - Oxford Univ. Press CY - Oxford ER - TY - THES A1 - Nooshiri, Nima T1 - Improvement of routine seismic source parameter estimation based on regional and teleseismic recordings T1 - Verbesserung von Standard-Verfahren zur Abschätzung von seismischen Quellparametern auf regionalen und teleseismischen Distanzen N2 - Seismological agencies play an important role in seismological research and seismic hazard mitigation by providing source parameters of seismic events (location, magnitude, mechanism), and keeping these data accessible in the long term. The availability of catalogues of seismic source parameters is an important requirement for the evaluation and mitigation of seismic hazards, and the catalogues are particularly valuable to the research community as they provide fundamental long-term data in the geophysical sciences. This work is well motivated by the rising demand for developing more robust and efficient methods for routine source parameter estimation, and ultimately generation of reliable catalogues of seismic source parameters. Specifically, the aim of this work is to develop some methods to determine hypocentre location and temporal evolution of seismic sources based on regional and teleseismic observations more accurately, and investigate the potential of these methods to be integrated in near real-time processing. To achieve this, a location method that considers several events simultaneously and improves the relative location accuracy among nearby events has been provided. This method tries to reduce the biasing effects of the lateral velocity heterogeneities (or equivalently to compensate for limitations and inaccuracies in the assumed velocity model) by calculating a set of timing corrections for each seismic station. The systematic errors introduced into the locations by the inaccuracies in the assumed velocity structure can be corrected without explicitly solving for a velocity model. Application to sets of multiple earthquakes in complex tectonic environments with strongly heterogeneous structure such as subduction zones and plate boundary region demonstrate that this relocation process significantly improves the hypocentre locations compared to standard locations. To meet the computational demands of this location process, a new open-source software package has been developed that allows for efficient relocation of large-scale multiple seismic events using arrival time data. Upon that, a flexible location framework is provided which can be tailored to various application cases on local, regional, and global scales. The latest version of the software distribution including source codes, a user guide, an example data set, and a change history, is freely available to the community. The developed relocation algorithm has been modified slightly and then its performance in a simulated near real-time processing has been evaluated. It has been demonstrated that applying the proposed technique significantly reduces the bias in routine locations and enhance the ability to locate the lower magnitude events using only regional arrival data. Finally, to return to emphasis on global seismic monitoring, an inversion framework has been developed to determine the seismic source time function through direct waveform fitting of teleseismic recordings. The inversion technique can be systematically applied to moderate- size seismic events and has the potential to be performed in near real-time applications. It is exemplified by application to an abnormal seismic event; the 2017 North Korean nuclear explosion. The presented work and application case studies in this dissertation represent the first step in an effort to establish a framework for automatic, routine generation of reliable catalogues of seismic event locations and source time functions. N2 - Seismologische Dienste spielen eine wichtige Rolle in der seismologischen Forschung und Gefährdungsanalyse, indem sie Quellparameter für seismische Ereignisse (Lokalisierung, Stärke, Mechanismus) bereitstellen und diese Daten langfristig verfügbar machen. Die Verfügbarkeit von Katalogen seismischer Quellparameter ist eine wichtige Voraussetzung für die Bewertung der seismischen Gefährdung und Minderung der Auswirkungen von Erdbeben. Die seismischen Kataloge sind für die Forschungsgemeinschaft besonders wertvoll, da sie grundlegende Hintergrundsdaten über lange Zeiträume liefern. Die vorliegende Arbeit ist motiviert durch die steigende Nachfrage nach der Entwicklung robusterer und effizienterer Methoden für die routinemäßige Schätzung von Quellparametern und schließlich die Verbesserung der Qualität und Quantität der seismischen Kataloge. Ziel dieser Arbeit ist es insbesondere Methoden zu entwickeln, um die Lokalisierung von Hypozentren und die zeitliche Entwicklung der Momentfreisetzung bei einem Erdbeben auf der Grundlage regionaler und teleseismischer Beobachtungen genauer zu bestimmen, und das Potenzial dieser Methoden für eine zeitnahe Verarbeitung zu untersuchen. Um dies zu erreichen, wurde eine Lokalisierungsmethode bereitgestellt, die mehrere Ereignisse gleichzeitig berücksichtigt und die relative Genauigkeit der Hypozentren zwischen benachbarten Ereignissen verbessert. Dieses Verfahren versucht, die Verzerrungen durch Geschwindigkeits-heterogenitäten zu reduzieren, oder äquivalent Einschränkungen und Ungenauigkeiten in dem angenommenen Geschwindigkeitsmodell auszugleichen, indem richtungsabhängige Zeitkorrekturen für jede seismische Station berechnet werden. Die systematischen Fehler, die durch die Ungenauigkeiten in der angenommenen Geschwindigkeitsstruktur in die Lokalisierung projiziert werden, können korrigiert werden, ohne explizit nach einem Geschwindigkeitsmodell zu lösen. Die Anwendung auf Ensemble von Erdbeben in tektonischen Situationen mit stark heterogener, komplexer Struktur wie Subduktionszonen und Plattengrenzregionen zeigt, dass diese Relokalisierung die Hypozentren im Vergleich zu Standardlokalisierungen signifikant verbessert. Um den rechnerischen Anforderungen der Lokalisierung gerecht zu werden, wurde ein neues Open-Source-Softwarepaket entwickelt, das eine effiziente Relokalisierung einer großen Anzahl von Erdbeben unter Verwendung von Ankunftszeitdaten ermöglicht. Daraufhin wird ein flexibler Lokalisierungsrahmen bereitgestellt, der an verschiedene Fälle auf lokaler, regionaler und globaler Ebene angepasst werden kann. Die neueste Version der Software einschließlich Quellcodes, Benutzerhandbuch, Beispieldatensatz und Änderungsverlauf ist für die Öffentlichkeit frei verfügbar. Der entwickelte Relokalisierungs-Algorithmus wurde geringfügig verändert und anschließend seine Leistung in einem simulierten Echtzeit Prozess bewertet. Es zeigte sich, dass die Anwendung der vorgeschlagenen Technik die Verzerrung der Hypozentren der Routineauswertung erheblich verringert und die Fähigkeit zum Lokalisieren der Ereignisse mit niedriger Magnitude unter Verwendung nur regionaler Ankunftsdaten verbessert. Um sich wieder auf die globale seismische Überwachung zu konzentrieren, wurde schließlich ein Inversionssystem entwickelt, um die Zeitfunktion der seismischen Quelle durch direkte Wellenformanpassung von teleseismischen Aufzeichnungen zu bestimmen. Die Inversionstechnik kann systematisch auf seismische Ereignisse mittlerer Größe angewendet werden und hat das Potenzial, in nahezu Echtzeitanwendungen ausgeführt zu werden. Dies wird beispielhaft durch die Anwendung auf ein abnormales seismisches Ereignis veranschaulicht: die nordkoreanische Atomexplosion 2017. Die in dieser Dissertation vorgestellten Arbeits- und Anwendungsfallstudien stellen den ersten Schritt dar, um einen Rahmen für die automatische, routinemäßige Erzeugung zuverlässiger Kataloge von seismischen Ereignisorten und Quellzeitfunktion zu schaffen. KW - observational seismology KW - earthquake bulletin data KW - seismic event localization KW - seismic source-time function estimation KW - beobachtende Seismologie KW - Erdbebenkatalogdaten KW - seismische Ereignislokalisierung KW - Herdzeit Parameter Abschätzung Y1 - 2020 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:kobv:517-opus4-459462 ER - TY - JOUR A1 - Nooshiri, Nima A1 - Bean, Christopher J. A1 - Dahm, Torsten A1 - Grigoli, Francesco A1 - Kristjansdottir, Sigriour A1 - Obermann, Anne A1 - Wiemer, Stefan T1 - A multibranch, multitarget neural network for rapid point-source inversion in a microseismic environment BT - examples from the Hengill Geothermal Field, Iceland JF - Geophysical journal international N2 - Despite advanced seismological techniques, automatic source characterization for microseismic earthquakes remains difficult and challenging since current inversion and modelling of high-frequency signals are complex and time consuming. For real-time applications such as induced seismicity monitoring, the application of standard methods is often not fast enough for true complete real-time information on seismic sources. In this paper, we present an alternative approach based on recent advances in deep learning for rapid source-parameter estimation of microseismic earthquakes. The seismic inversion is represented in compact form by two convolutional neural networks, with individual feature extraction, and a fully connected neural network, for feature aggregation, to simultaneously obtain full moment tensor and spatial location of microseismic sources. Specifically, a multibranch neural network algorithm is trained to encapsulate the information about the relationship between seismic waveforms and underlying point-source mechanisms and locations. The learning-based model allows rapid inversion (within a fraction of second) once input data are available. A key advantage of the algorithm is that it can be trained using synthetic seismic data only, so it is directly applicable to scenarios where there are insufficient real data for training. Moreover, we find that the method is robust with respect to perturbations such as observational noise and data incompleteness (missing stations). We apply the new approach on synthesized and example recorded small magnitude (M <= 1.6) earthquakes at the Hellisheioi geothermal field in the Hengill area, Iceland. For the examined events, the model achieves excellent performance and shows very good agreement with the inverted solutions determined through standard methodology. In this study, we seek to demonstrate that this approach is viable for microseismicity real-time estimation of source parameters and can be integrated into advanced decision-support tools for controlling induced seismicity. KW - Neural networks KW - fuzzy logic KW - Computational seismology KW - Induced seismicity KW - Earthquake source observations Y1 - 2021 U6 - https://doi.org/10.1093/gji/ggab511 SN - 0956-540X SN - 1365-246X VL - 229 IS - 2 SP - 999 EP - 1016 PB - Oxford Univ. Press CY - Oxford ER -