Terms, citation and credits

Version of 11 October 2026.

What these tools are

SeisForge tools are exploratory research software from the Meng group at UCLA. They are not peer reviewed (unless a paper is cited on the tool page), come with no warranty, and must not be used for operational, safety or engineering decisions. Every set, score and uncertainty is stated against the reference you supply (for example another catalogue), not against physical truth, and its guarantee holds only under the exchangeability condition described on the tool page.

Use

Academic and non-commercial research use is free. No account is needed.
Commercial use needs a separate written permission: this includes using the service or its results in a commercial product, a paid service or consulting work. Please write to Lingsen Meng, [email protected].
The software itself is not released. No licence to copy or redistribute the code is granted by using the service. For reuse of the code, write to the address above.
Automated or bulk use is limited by the rate limits stated on the tool page.

Your data

Uploaded text is processed in memory by a worker on a research workstation and is not written to disk by the service. Results are kept in memory for one hour and then deleted. The service keeps no accounts and no copy of your files. Rate limiting uses your network address for up to one hour. Please do not upload personal or confidential data. You keep all rights to your data and results.

If you publish results obtained with the service

Please cite the SeisForge paper and the module-specific references below, and state the service address and the date of use.

SeisForge: Meng, L., Huang, H., Ma, J., & Ma, Y. (2026). Can human seismologists verify machine-learning detected earthquakes? EarthArXiv. doi:10.31223/X5MZ2D (version 3; replace by the journal reference when published).
Detection module (confirmability score): Meng, Huang, Ma & Ma (2026) above. Example catalogue: 2019 Ridgecrest machine-learning catalogue of the study, Zenodo doi:10.5281/zenodo.22059218, CC BY 4.0; reference catalogue: Shelly, D. R. (2020), Seismol. Res. Lett. 91, 1971–1978, doi:10.1785/0220190309, and the SCSN analyst catalogue (SCEDC doi:10.7909/C3WD3xH1). The page scores in your browser and uploads nothing.
Focal-mechanism module: methods paper in preparation; until it appears, cite the SeisForge paper above and this service (SeisForge, Focal mechanisms, seisforge.imageolab.com/focal, accessed date).
Stress inversion (when the stress option is used): Vavryčuk, V. (2014). Iterative joint inversion for stress and fault orientations from focal mechanisms. Geophys. J. Int., 199, 69–77. doi:10.1093/gji/ggu224. The service runs the STRESSINVERSE code of V. Vavryčuk for research purposes; it is not offered for download here.
Stress paper (to be written): placeholder, will be added when the preprint is posted.

Data and methods behind the examples

Amatrice–Norcia mechanisms and polarity inputs: Song, X. (2025), The Training, Test, and Evaluation sets for FocoNet v1.0, Zenodo, doi:10.5281/zenodo.17822343, CC BY 4.0 (derived values only; no model weights shipped). Model: Song, Meier, Ellsworth & Beroza (2026), FocoNet: Transformer-Based Focal-Mechanism Determination, JGR Machine Learning and Computation 3, doi:10.1029/2025JH000879.
Southern California polarities and catalogue labels: Southern California Seismic Network, doi:10.7914/SN/CI, and SCEDC, doi:10.7909/C3WD3xH1. Classifier: Zhao, Xiao, Zhang & Yang (2023), DiTingMotion, Front. Earth Sci. 11, doi:10.3389/feart.2023.1103914.
California focal-mechanism catalogue and stress model (original versus refined mechanisms): Cheng, Y., Bürgmann, R., Taira, T., Liao, Z., & Allen, R. (2025), Statewide California focal mechanism catalog and stress model (1981–2021), Mendeley Data v2, doi:10.17632/2np9vw5v7w.2, CC BY 4.0; and Cheng, Y., Bürgmann, R., Taira, T., Liao, Z., & Allen, R. M. (2026), Stress Model of California: Fault-Stress Interactions Across a Complex Plate Boundary System From Focal Mechanisms of Small Earthquakes, J. Geophys. Res. Solid Earth 131, doi:10.1029/2025JB032164.
Ridgecrest 2019: Cheng, Y. & Ben-Zion, Y. (2020), A focal mechanism catalog for Ridgecrest sequence derived with deep learning algorithms, Mendeley Data v1, doi:10.17632/h6rbr6d6n2.1; Cheng, Y. (2021), moment tensor solutions of events in the 2019 Ridgecrest earthquake sequence, Mendeley Data v1, doi:10.17632/6p3dgbrptv.1; Atterholt, Wilding & Ross (2025), Geophys. J. Int. 240, 1579–1592, doi:10.1093/gji/ggaf001 (data doi:10.22002/5af05-cah73).
Michael, A. J. (1984), Determination of stress from slip data, J. Geophys. Res. 89, 11517–11526, doi:10.1029/JB089iB13p11517.

Questions, corrections and permissions: [email protected].