Phase Picking¶
The FAST output is interfaced with SeisBench to pick the arrival time of phases automatically with machine learning models.
SeisBench is an open-source python toolbox for machine learning in seismology. Learn more about SeisBench here.
Cut SAC Files¶
In cut_event_files.py
, change the station list on line 14 to list the stations in your dataset and the name of the final detection list on line 16.
Example in cut_event_files
:
Your continuous seismic data should be found in /FAST/data/waveforms*/Deci5.Pick.*
Pick Phases Over All Input¶
Note
If stations in your dataset do not have 3 channels, you will need to alter run_seisbench.py
to make .sac file copies for stations without 3 channels.
In the Hector Mine dataset, only the HEC station has 3 channels. On line 20 of run_seisbench.py
, the stations that only have one component are in a list, which is iterated through to make channel copies for phase picking.
To make sure run_seisbench.py
runs correctly on your dataset, edit the following code in the script to account for missing station channels:
Run the SeisBench script:
The plotted annotations with each station's waveform and EQTransformer's picks and detections for every event is found in:
Example annotated plot from one event in the Hector Mine dataset:
The phase pick information found from SeisBench is found in:
Example output:
- "peak_time": Arrival time of pick
- "peak_value": Probability of pick