English
 
Privacy Policy Disclaimer
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT
Add to Basket
  Local TagsRelease HistoryDetailsSummary

Released
Conference Paper

Using deep learning and spectrograms to identify anthropogenic signals in an urban setting

Authors

Velasco,  Aaron A.
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

Ayala Cortez,  Solyma
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

Garcia,  Marc
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

Marianne,  Karplus
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

April,  Leah
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

External Ressource
No external resources are shared
Fulltext (public)
There are no public fulltexts stored in GFZpublic
Supplementary Material (public)
There is no public supplementary material available
Citation

Velasco, A. A., Ayala Cortez, S., Garcia, M., Marianne, K., April, L. (2023): Using deep learning and spectrograms to identify anthropogenic signals in an urban setting, XXVIII General Assembly of the International Union of Geodesy and Geophysics (IUGG) (Berlin 2023).
https://doi.org/10.57757/IUGG23-4743


Cite as: https://gfzpublic.gfz-potsdam.de/pubman/item/item_5021151
Abstract
El Paso, Texas, USA, along with its sister city Ciudad Juarez, Mexico, is the second largest binational city worldwide, and lies in a tectonically-active rift zone: the Rio Grande Rift. The rate of rifting is relatively slow, ~2 mm/year, but historic earthquakes suggest the potential of large earthquakes in the area. From June 5-14, 2021, a 50 Magseis Fairfield Z-Land 5-Hz 3-component nodal seismic network was deployed to record seismic data in El Paso, Texas in order to characterize local site response in an urban setting. We utilize data from this local network to create a seismic event catalog and attempt to differentiate anthropogenic sources from naturally-occurring ones using a suite of approaches, including machine learning, time-frequency analysis, beamforming, power spectral density, and cross-correlation. We first use a deep neural network-based seismic algorithm to identify seismic arrivals from natural and possibly mining events. We then will run an optimized STA/LTA detector to identify emergent, long duration signals, typical of trains, and then compare our results to the train schedule, where train traffic is very active. Trains can also be identified using spectral and beamforming analysis. We will use these techniques to compare our results with those from the STA/LTA and machine learning approaches. Finally, we will compare the differences in energy recorded by stations with variable site surface geology, incorporating signals recorded by a permanent broadband station as a reference.