Experiments using the Triggered Tremor dataset.
The seismic data for this project was provided by a number of different sources. The Broadband Array in Taiwan for Seismology and the Central Weather Bureau of Taiwan provided seismograms containing tremor that was dynamically triggered by seismic surface waves from one of six large earthquakes. The earthquakes are:
Manitude | Location | Datetime |
---|---|---|
M7.8 | Southern Qinghai, China | 2001-11-14 09:26:10 (UTC) |
M8.2 | Hokkaido, Japan region | 2003-09-25 19:50:06 (UTC) |
M9.1 | 2004 Sumatra - Andaman Islands Earthquake | 2004-12-26 00:58:53 (UTC) |
M8.6 | Northern Sumatra, Indonesia | 2005-03-28 16:09:36 (UTC) |
M8.4 | Northern Sumatra, Indonesia | 2007-09-12 11:10:26 (UTC) |
M8.2 | Hokkaido, Japan region | 2003-09-25 19:50:06 (UTC) |
M9.1 | 2011 Great Tohoku Earthquake, Japan | 2011-03-11 05:46:24 (UTC) |
The High Sensitivity Seismograph Network Japan provided seismograms containing tremor that was dynamically triggered by seismic surface waves from the following earthquake.
Manitude | Location | Datetime |
---|---|---|
M8.6 | Off the west coast of northern Sumatra | 2012-04-11 08:38:36 (UTC) |
The positive identifications of dynamically triggered tremor were made by Kevin Chao and Vivian Tang. These seismograms are called "positive examples". All positive examples begin D/4
s after the earthquake's origin time, where D
is the distance in km from the epicenter to the recording station, and they end 1000
s later.
Negative examples consist of seismograms of the same duration (1000
s) recorded on 2011-03-04
, a day without large earthquakes by stations from the following networks:
- Broadband Array in Taiwan for Seismology
- Australian National Seismograph Network
- Global Seismograph Network: UI & II
- Malaysian National Seismic Network
Each positive or negative example is one component of a three-component recording. Each example was band-pass filtered with corner frequencies of 2
and 5
Hz.
Data is made public:
- Raw Data
- above pre-processing is already applied
- Model-ready data can be reproduced by simply combining all files into a single directory and noting the label
- no additonal pre-processing unless noted in the experiment readme.
- Model-ready datasets are detailed internally in seisml