Subtopic Deep Dive
Seismic Phase Picking
Research Guide
What is Seismic Phase Picking?
Seismic phase picking is the automated detection and precise estimation of P- and S-wave arrival times in seismograms using deep learning models such as convolutional neural networks.
Researchers apply models like PhaseNet (Zhu and Beroza, 2018, 996 citations) and Earthquake Transformer (Mousavi et al., 2020, 952 citations) to handle noisy data from growing seismic networks. These methods outperform traditional approaches by processing continuous waveforms for simultaneous detection and picking. Benchmark datasets like STEAD (Mousavi et al., 2019, 407 citations) support model training and evaluation.
Why It Matters
Accurate phase picking enables rapid earthquake location, magnitude estimation, and early warning systems, critical for seismic hazard mitigation. PhaseNet by Zhu and Beroza (2018) processes large datasets from global networks, improving monitoring efficiency. Earthquake Transformer by Mousavi et al. (2020) handles microearthquakes in noisy environments, enhancing real-time event catalogs used by agencies like USGS.
Key Research Challenges
Noisy Data Handling
Seismic records often contain clutter from cultural noise and scattered waves, reducing picking accuracy. Mousavi et al. (2020) address this with attention mechanisms in Earthquake Transformer (952 citations). Traditional filters struggle with variable signal-to-noise ratios.
Phase Association
Linking detected P- and S-phases to specific earthquakes requires robust algorithms amid overlapping events. Ross et al. (2018) use deep learning for generalized detection (477 citations), but association errors persist in dense swarms. Benchmarking needs standardized metrics.
Dataset Scalability
Training data shortages limit model generalization across regions and magnitudes. STEAD dataset by Mousavi et al. (2019, 407 citations) provides global signals, yet labeled data remains costly to curate. Transfer learning helps but requires diverse sources.
Essential Papers
PhaseNet: A Deep-Neural-Network-Based Seismic Arrival Time Picking Method
Weiqiang Zhu, Gregory C. Beroza · 2018 · Geophysical Journal International · 996 citations
As the number of seismic sensors grows, it is becoming increasingly difficult for analysts to pick seismic phases manually and comprehensively, yet such efforts are fundamental to earthquake monito...
Earthquake transformer—an attentive deep-learning model for simultaneous earthquake detection and phase picking
S. Mostafa Mousavi, William L. Ellsworth, Weiqiang Zhu et al. · 2020 · Nature Communications · 952 citations
Abstract Earthquake signal detection and seismic phase picking are challenging tasks in the processing of noisy data and the monitoring of microearthquakes. Here we present a global deep-learning m...
Imaging dispersion curves of surface waves on multi‐channel record
Choon Byong Park, Richard D. Miller, Jianghai Xia · 1998 · 762 citations
PreviousNext No AccessSEG Technical Program Expanded Abstracts 1998Imaging dispersion curves of surface waves on multi‐channel recordAuthors: Choon Byong ParkRichard D. MillerJianghai XiaChoon Byon...
Earthquake research in China
Anonymous · 1975 · Eos · 576 citations
A visit to China of an American seismological delegation, which took place October 5 to November 5, 1974, is covered in this report. The Americans were sponsored by the Committee on Scholarly Commu...
Guidelines for the good practice of surface wave analysis: a product of the InterPACIFIC project
Sebastiano Foti, Fabrice Hollender, Flora Garofalo et al. · 2017 · Bulletin of Earthquake Engineering · 516 citations
Generalized Seismic Phase Detection with Deep Learning
Zachary E. Ross, Men‐Andrin Meier, Egill Hauksson et al. · 2018 · Bulletin of the Seismological Society of America · 477 citations
To optimally monitor earthquake-generating processes, seismologists have\nsought to lower detection sensitivities ever since instrumental seismic\nnetworks were started about a century ago. Recentl...
STanford EArthquake Dataset (STEAD): A Global Data Set of Seismic Signals for AI
S. Mostafa Mousavi, Yixiao Sheng, Weiqiang Zhu et al. · 2019 · IEEE Access · 407 citations
Seismology is a data rich and data-driven science. Application of machine learning for gaining new insights from seismic data is a rapidly evolving sub-field of seismology. The availability of a la...
Reading Guide
Foundational Papers
Start with PhaseNet (Zhu and Beroza, 2018) for core CNN picking method, then Generalized Seismic Phase Detection (Ross et al., 2018) for waveform scanning basics.
Recent Advances
Study Earthquake Transformer (Mousavi et al., 2020) for attention-based advances and STEAD (Mousavi et al., 2019) for dataset standards.
Core Methods
Core techniques include CNNs (PhaseNet), U-Nets with attention (Earthquake Transformer), and hybrid CNN-RNN (CRED). Probability waveforms and pick association follow.
How PapersFlow Helps You Research Seismic Phase Picking
Discover & Search
Research Agent uses searchPapers to find PhaseNet (Zhu and Beroza, 2018), then citationGraph reveals 996 citing works including Earthquake Transformer, and findSimilarPapers uncovers STEAD (Mousavi et al., 2019) for datasets. exaSearch queries 'seismic phase picking benchmarks' to surface Ross et al. (2018).
Analyze & Verify
Analysis Agent applies readPaperContent to extract PhaseNet architecture details, verifyResponse with CoVe checks claims against STEAD waveforms, and runPythonAnalysis replots seismograms with NumPy for pick error stats. GRADE grading scores model performance evidence from Mousavi et al. (2020).
Synthesize & Write
Synthesis Agent detects gaps like regional dataset biases via contradiction flagging across Zhu (2018) and Ross (2018), then Writing Agent uses latexEditText for methods sections, latexSyncCitations for 20+ references, and latexCompile for full reports. exportMermaid visualizes PhaseNet vs. Transformer architectures.
Use Cases
"Compute picking precision of PhaseNet on STEAD noisy waveforms"
Research Agent → searchPapers(PhaseNet) → Analysis Agent → readPaperContent + runPythonAnalysis(NumPy waveform simulation, error metrics) → researcher gets precision-recall curve plot and stats table.
"Draft LaTeX review comparing PhaseNet and Earthquake Transformer"
Synthesis Agent → gap detection → Writing Agent → latexEditText(draft) → latexSyncCitations(10 papers) → latexCompile → researcher gets compiled PDF with figures and bibliography.
"Find GitHub repos for CRED earthquake detector implementation"
Research Agent → searchPapers(CRED Mousavi 2019) → Code Discovery → paperExtractUrls → paperFindGithubRepo → githubRepoInspect → researcher gets code snippets, training scripts, and STEAD loader.
Automated Workflows
Deep Research workflow scans 50+ phase picking papers via searchPapers chains, producing structured reports ranking models by citation impact like Zhu (2018). DeepScan applies 7-step CoVe verification to benchmark PhaseNet against STEAD, with GRADE checkpoints. Theorizer generates hypotheses on attention mechanisms from Mousavi et al. (2020) for low-SNR picking.
Frequently Asked Questions
What is seismic phase picking?
Seismic phase picking identifies P- and S-wave arrival times in seismograms. Deep learning models like PhaseNet (Zhu and Beroza, 2018) automate this for large-scale monitoring.
What are key methods in seismic phase picking?
PhaseNet uses convolutional neural networks for pick probability waveforms (Zhu and Beroza, 2018). Earthquake Transformer employs attention for simultaneous detection and picking (Mousavi et al., 2020). CRED combines residuals and recurrent units (Mousavi et al., 2019).
What are the most cited papers?
PhaseNet (Zhu and Beroza, 2018, 996 citations), Earthquake Transformer (Mousavi et al., 2020, 952 citations), and Generalized Detection (Ross et al., 2018, 477 citations) lead citations.
What open problems exist?
Challenges include phase association in swarms, generalization to new regions, and real-time low-SNR picking. Datasets like STEAD help but lack labels for rare events.
Research Seismology and Earthquake Studies with AI
PapersFlow provides specialized AI tools for Computer Science researchers. Here are the most relevant for this topic:
AI Literature Review
Automate paper discovery and synthesis across 474M+ papers
Code & Data Discovery
Find datasets, code repositories, and computational tools
Deep Research Reports
Multi-source evidence synthesis with counter-evidence
AI Academic Writing
Write research papers with AI assistance and LaTeX support
See how researchers in Computer Science & AI use PapersFlow
Field-specific workflows, example queries, and use cases.
Start Researching Seismic Phase Picking with AI
Search 474M+ papers, run AI-powered literature reviews, and write with integrated citations — all in one workspace.
See how PapersFlow works for Computer Science researchers
Part of the Seismology and Earthquake Studies Research Guide