Deep Learning Applications in Seismic Event Detection

Recent advances in deep learning have revolutionised the detection and characterisation of seismic events by enabling automated, high-sensitivity analysis of continuous waveform data. Models built on convolutional and recurrent architectures, often augmented by attention mechanisms, demonstrate robust performance even in high-noise environments and across diverse tectonic settings. These approaches span both supervised and unsupervised frameworks, facilitating the identification of microseismic activity, accurate phase picking, and rapid event location. Practical implementations leverage global and regional datasets alongside transfer learning to bridge scale disparities, while unsupervised clustering methods offer novel insights into emergent seismic patterns. The integration of real-time processing pipelines and distributed sensing networks underscores the global significance of deep learning in seismic monitoring, promising enhanced hazard assessment and resource management applications.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

A comprehensive review of machine learning in microseismic monitoring highlights the transition from traditional signal processing to deep learning-based detection and hypocentre estimation. Convolutional and recurrent networks are shown to enhance event detection sensitivity, reduce false positives, and enable real-time monitoring when applied to distributed acoustic sensing data. The study emphasises the importance of uncertainty quantification and statistical by-products that improve event characterisation.

A large-scale benchmark comparing multiple deep learning models on diverse seismic datasets reveals that modern architectures, including transformer-inspired and residual networks, achieve human-level performance in event detection, phase identification, and onset picking. Cross-domain tests indicate that models trained on regional data generalise effectively to other regions of similar scale, while transfer to teleseismic distances remains challenging. An open framework supports ongoing model evaluation and deployment.

A transfer learning approach bridges the gap between laboratory-scale and field-scale seismic experiments by retraining an existing phase picker with a small subset of local seismogram data. The adapted model matches expert picks while reducing manual effort by orders of magnitude. Integration with double-difference tomography workflows yields improved event locations, demonstrating scalability across spatial and temporal domains.

Deep Learning Applications in Seismic Event Detection publication trend

The graph below shows the total number of articles in deep learning applications in seismic event detection across all publications each year (not limited to Nature Index journals).

Technical terms

Deep learning: A subset of machine learning employing multi-layer neural networks to learn hierarchical representations directly from data.

Convolutional neural network (CNN): A deep learning architecture that applies convolutional filters to extract spatial and temporal features from waveform inputs.

Attention mechanism: A network component that dynamically weights input features to focus on salient portions of waveform data, improving simultaneous detection and phase picking.

Phase picking: The process of identifying the arrival times of primary (P) and secondary (S) seismic waves on continuous recordings.

Microseismic monitoring: The detection and analysis of low-magnitude seismic events, often induced by industrial or natural processes, using dense sensor arrays and signal enhancement techniques.

References

1. Machine learning in microseismic monitoring. Earth-Science Reviews (2023).
2. Earthquake transformer—an attentive deep-learning model for simultaneous earthquake detection and phase picking. Nature Communications (2020).
3. CRED: A Deep Residual Network of Convolutional and Recurrent Units for Earthquake Signal Detection. Scientific Reports (2019).
4. STanford EArthquake Dataset (STEAD): A Global Data Set of Seismic Signals for AI. IEEE Access (2019).
5. Clustering earthquake signals and background noises in continuous seismic data with unsupervised deep learning. Nature Communications (2020).
6. Locating induced earthquakes with a network of seismic stations in Oklahoma via a deep learning method. Scientific Reports (2020).
7. Using a Deep Neural Network and Transfer Learning to Bridge Scales for Seismic Phase Picking. Geophysical Research Letters (2020).
8. Which Picker Fits My Data? A Quantitative Evaluation of Deep Learning Based Seismic Pickers. Journal of Geophysical Research: Solid Earth (2022).