Skip to content

quapity/Utah

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

51 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Binary Event Classification on UUSS Data

Overview | Data | Models | Credits

Overview

The University of Utah Seismograph Stations (UUSS) has been monitoring seismicity in Utah for decades. This work uses just a few years (2012-2017) of the UUSS seismicity catalog to test how well deep learning can assign event type using waveform data from individual stations. Although many event types are observed during routine processing, discrimination of quarry blasts and earthquakes can be among the most time consuming. This work focuses on binary event discrimination for quarry blasts and earthquakes at local distances (< 350 km).

What we found is that we are able to match event types in Utah very well using a variety of architectures when we rely on event spectrograms (90 seconds in duration). Each quarry site produces many examples, and even though examples from an individual mine site can vary significantly, identifying new examples from these mine sites is successful. Additionally, we were able to make successful predictions using both vertical and triaxial sensors using one model, which makes this a practical, flexible and often highly accurate alternative or compliment to waveform matching methods.

Data

In data directory includes a csv of the catalog metadata we used. This does not include the data arrays, but gives users the first arrival pick times for stations and includes some event data (mag, type, etc). Using obspy users can download data arrays for use in whatever format is desirable. Alternative formats and additional Utah event catalogs through 2020 can be found at https://doi.org/10.31905/RDQW00CT.

Models

The LSTM models we used in this study were built with Keras and are provided here. Subsequent studies that leverage this dataset use pytorch, for example semi-supervised learning for event discrimination.

Credits

Reference:

Linville, L., Pankow, K., & Draelos, T. (2019). Deep learning models augment analyst decisions for event discrimination. Geophysical Research Letters, 46(7), 3643-3651.

About

model notes for UUSS event classification

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published