Data-driven exploration of continuous seismograms with unsupervised learning
| Autor: | Steinmann, R., Seydoux, L., Campillo, M. |
|---|---|
| Rok vydání: | 2023 |
| Zdroj: | XXVIII General Assembly of the International Union of Geodesy and Geophysics (IUGG) |
| DOI: | 10.57757/iugg23-3388 |
| Popis: | Continuous seismograms are considered a goldmine of information, however, the complexity and size of seismic data challenge the efficient and successful mining of interesting information. Machine learning methods scanning continuous data streams can help overcome these challenges and might reveal new types of seismic signals, offering new insights about active geological objects. We present two study cases where a scattering network transforms continuous single-station seismograms into a stable data representation, serving as a basis for further data exploration. In the first case, an independent component analysis picks up blindly the continuous medium change due to freezing and thawing in an urban and noisy environment. In the second case, manifold learning techniques reveal the ever-changing and non-stationary nature of volcanic tremors recorded at the Klyuchevskoy volcano, Russia, across a time scale of one year. Both examples confirm that unsupervised learning can help decipher seismic recordings' complex continuous evolution, containing relevant information about the subsurface. The 28th IUGG General Assembly (IUGG2023) (Berlin 2023) |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|