AI‐PAL: Self‐Supervised AI Phase Picking via Rule‐Based Algorithm for Generalized Earthquake Detection

Delineating fault structures through microseismicity is crucial for earthquake hazard assessment, yet constructing high‐resolution catalogs over extended periods remains challenging. This study introduces AI‐PAL, a novel deep learning‐driven workflow that employs a Self‐Attention RNN (SAR) model tra...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Journal of geophysical research. Solid earth Ročník 130; číslo 4
Hlavní autori: Zhou, Yijian, Ding, Hongyang, Ghosh, Abhijit, Ge, Zengxi
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Washington Blackwell Publishing Ltd 01.04.2025
Predmet:
Algorithms
Artificial intelligence
Catalogues
data collection
Deep learning
Earthquakes
east anatolian fault zone
Fault lines
fault structure
Fault zones
generalizability
Geological faults
Geological hazards
geophysics
Hazard assessment
hazard characterization
Information processing
Labels
Machine learning
Matched filters
Neural networks
ridgecrest‐coso
Scientists
Seismic activity
seismic catalog
Seismic data
Seismic hazard
Seismological data
Structures
Supervised learning
Template matching
Training
Turkey (country)
Workflow
Turkey
ISSN:2169-9313, 2169-9356
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:Delineating fault structures through microseismicity is crucial for earthquake hazard assessment, yet constructing high‐resolution catalogs over extended periods remains challenging. This study introduces AI‐PAL, a novel deep learning‐driven workflow that employs a Self‐Attention RNN (SAR) model trained with detections from PAL, an established rule‐based algorithm (Zhou, Yue, et al., 2021, https://doi.org/10.1785/0220210111), for generalized earthquake detection. PAL utilizes short‐term‐average over long‐term‐average algorithm for event detection, ensuring consistent performance across different datasets. AI‐PAL leverages these rule‐based picks as training labels, enabling self‐supervised learning of the SAR model across arbitrary regions, thereby enhancing PAL's detection capabilities. We applied SAR‐PAL to two distinct regions that are featured by recent large earthquakes: (a) the preseismic period of the Ridgecrest‐Coso region (2008–2019), and (b) the pre‐to‐postseismic period of the East Anatolian Fault Zone (EAFZ, 2020–2023/04). Our results demonstrate that SAR‐PAL offers slightly higher detection completeness than the quake template matching matched filter catalog, while boosts over 100 times faster processing and a superior temporal stability, avoiding detection gaps during background periods. Compared to PhaseNet and GaMMA, two widely recognized phase picker and associator, SAR‐PAL proved more scalable, achieving ∼2.5 times more event detections in the EAFZ case, along with a ∼7 times higher phase association rate. We further experimented training PhaseNet and SAR with PAL detections and routine catalogs, and found that no other combinations matched the detection performance of SAR‐PAL. The enhanced catalogs built by SAR‐PAL reveals geometrical complexities of the Ridgecrest faults and the Erkenek‐Pütürge segment of EAFZ, offering insights into their contrasting roles during the large earthquake. Plain Language Summary Seismic monitoring helps scientists understand fault structures and assess earthquake hazards, but building detailed earthquake catalogs over long periods remains difficult. Our study introduces AI‐PAL, a new deep learning approach that improves earthquake detection by combining artificial intelligence with rule‐based methods. AI‐PAL uses an existing algorithm, PAL, to automatically generate training labels, allowing a neural network to learn from regional seismic data without requiring manually labeled events. This self‐supervised learning approach enables more accurate and efficient detection of small earthquakes. We tested AI‐PAL in two earthquake‐prone regions: the Ridgecrest‐Coso area in California and the East Anatolian Fault Zone in Turkey. Our results show that AI‐PAL detects more earthquakes than traditional methods while being 100 times faster and more reliable over time. Compared to other AI‐based seismic tools, AI‐PAL identified 2.5 times more earthquakes and associated seismic phases with 7 times greater accuracy in Turkey. These improvements allow scientists to build more complete earthquake catalogs, revealing fault structures that were previously undetectable. By applying AI‐PAL, researchers can better monitor earthquake activity and improve hazard assessments, ultimately supporting safer communities in seismically active regions. Key Points We developed AI‐PAL, a catalog construction workflow that trains an artificial intelligence (AI) phase picker with PAL detections, an established rule‐based algorithm AI‐PAL demonstrated greater temporal stability than matched filter, while also offering improved scalability compared to pretrained models The performance of AI picker is influenced by the quality of training data, with different models adapting to various sampling strategies
Bibliografia:This article was corrected on 10 APR 2025. See the end of the full text for details.
ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ISSN:2169-9313
2169-9356
DOI:10.1029/2025JB031294