English · 简体中文
AI for Seismology · Vibe Seismology
Seismic waveforms · Foundation models · Scientific agents
Seismic Foundation Models · AI for Research · Automated Cataloging · Courses & Tutorials
My focus is AI for Seismology: building models, tools, and agents that help us understand seismic signals and carry out earthquake research. I also explore Vibe Seismology — working with AI through natural language to turn research ideas into code, analyses, and reproducible workflows.
- Signals & models: seismic waveform representations, multitask learning, phase picking, and event identification.
- Analysis & inversion: earthquake location, surface-wave dispersion inversion, numerical computing, and GPU acceleration.
- Vibe Seismology: natural language interaction, RAG, tool calling, and scientific agents for automated cataloging and reproducible analysis.
| Project | What it does |
|---|---|
| SeismicXM | A cross-task foundation model for single-station seismic waveforms, supporting phase picking, first-motion polarity classification, and event-type classification. |
| SAGE | An AI workbench for seismology research that connects natural language interaction, knowledge retrieval, code execution, scientific plotting, and paper writing. |
| SeismicX Agent | A seismic monitoring and analysis platform combining real-time data acquisition, deep learning phase detection, and event association parameter tuning. |
| Seismological AI Tools | A collection of AI tools for seismological research. |
| CSNBench | A benchmark comparing seismic phase-picking models in China. |
| SeismicX Catalog Skill | Tools for AI agents to build earthquake catalogs from continuous seismic waveforms. |
More projects for datasets, location, inversion, and scientific computing:
- seis-stream — A continuous seismic waveform dataset for benchmarking earthquake monitoring algorithms.
- bayes_location — Travel-time models, robust Bayesian earthquake location, and catalog quality control.
- SurfFlow — Probabilistic surface-wave dispersion inversion using a conditional rectified flow.
- grtm_cuda — GPU-accelerated Green's function computation for layered media.
I also share resources that connect theory with working code:
- Quantitative Seismology · Learning materials in Chinese
- Deep Learning Theory & Practice · Companion code
- Large Language Model Course
- RAG, Tool Calling & Agent Development
I welcome discussions about AI for Seismology, Vibe Seismology, and research tools. For project-specific questions, please open an issue in the relevant repository. For academic use or commercial collaboration, please review each project's license and usage guidelines.
📬 Email: yuziye@cea-igp.ac.cn
Explore signals. Build models. Make research reproducible.