EQCCT package is a production-ready EarthQuake detection and phase-picking method using the Compact Convolutional Transformer
If you find this package useful, please do not forget to cite the following paper.
Saad, O.M., Chen, Y.F., Siervo, D., Zhang, F., Savvaidis, A., Huang, G., Igonin, N., Fomel, S., and Chen, Y., (2023). EQCCT: A production-ready EarthQuake detection and phase picking method using the Compact Convolutional Transformer, IEEE Transactions on Geoscience and Remote Sensing, 61, 4507015, doi:10.1109/TGRS.2023.3319440.
Chen, Y., Savvaidis, A., Saad, O.M., Siervo, D., Huang, G., Chen, Y.F., Grigoratos, I., Fomel, S., and Breton, C., (2024). Thousands of Induced Earthquakes per Month in West Texas Detected Using EQCCT, Geosciences, 14(5), 114, doi: 10.3390/geosciences14050114.
BibTeX:
@article{eqcct,
author={Omar M. Saad and Yunfeng Chen and Daniel Siervo and Fangxue Zhang and Alexandros Savvaidis and Guo-chin Huang and Nadine Igonin and Sergey Fomel and Yangkang Chen},
title = {EQCCT: A production-ready EarthQuake detection and phase picking method using the Compact Convolutional Transformer},
journal={IEEE Transactions on Geoscience and Remote Sensing},
year=2023,
volume=61,
issue=12,
pages={4507015},
doi={10.1109/TGRS.2023.3319440},
}
@Article{eqcct2024catalog,
author={Yangkang Chen and Alexandros Savvaidis and Omar M. Saad and Daniel Siervo and Guo-chin Dino Huang and Yunfeng Chen and Iason Grigoratos and Sergey Fomel and Caroline Breton},
title = {Thousands of Induced Earthquakes per Month in West Texas Detected Using EQCCT},
journal={Geosciences},
year=2024,
volume=14,
issue=5,
number=5,
pages={114},
doi={10.3390/geosciences14050114},
}
The accepted version of the paper can be downloaded from:
https://drive.google.com/drive/folders/1v9akXtnNy6b3gCWzUn9wTBiAsKPgHlKM?usp=sharing
Developers of the EQCCT package, 2021-present
MIT License
First set up the environment and install the dependency packages
conda create -n eqcct python=3.7.16
conda activate eqcct
conda install ipython notebook
pip install obspy tqdm matplotlib-scalebar tensorflow==2.8.0 protobuf==3.20.1 pandas==1.3.5 scikit-learn==1.0.2
Then install EQCCT using the latest version
git clone https://github.com/chenyk1990/eqcct
cd eqcct
pip install -v -e .
Or using Pypi
pip install eqcct
Or using pip directly from Github
pip install git+https://github.com/chenyk1990/eqcct
1 Pick the P/S arrivals from an arbitrary waveform (stream) containing an earthquake
import obspy.core.utcdatetime as utc
from obspy.clients.fdsn import Client
from eqcct.stream import st_predictor
from eqcct import plot_traces
import pandas as pd
import os
os.system('rm -rf ./detections_ALPN')
cl=Client('http://rtserve.beg.utexas.edu')
st = cl.get_waveforms('TX', 'ALPN', '00', '*', utc.UTCDateTime(2022, 12, 16, 23, 35, 27, 000000), utc.UTCDateTime(2022, 12, 16, 23, 37, 27, 000000))
st_predictor(input_modelP='../ModelPS/test_trainer_024.h5',
input_modelS='../ModelPS/test_trainer_021.h5',
stinput = st,
output_dir='detections_ALPN',
P_threshold=0.1,
S_threshold=0.1,
number_of_plots=10,
normalization_mode='std',
batch_size=1,
overlap=0.8,
gpuid=None,
gpu_limit=None)
data = pd.read_csv('detections_ALPN/ALPN_outputs/X_prediction_results.csv', low_memory=False)
data.p_arrival_time
pat=[ii for ii in list(data.p_arrival_time) if type(ii) is str]; #P arrival time
sat=[ii for ii in list(data.s_arrival_time) if type(ii) is str]; #S arrival time
import obspy
plot_traces(st,axoff=1,titleoff=1,ptime= [obspy.UTCDateTime(pat[0]),obspy.UTCDateTime(pat[0]),obspy.UTCDateTime(pat[0])],stime=[obspy.UTCDateTime(sat[0]),obspy.UTCDateTime(sat[0]),obspy.UTCDateTime(sat[0])],figname='test_eqcct_texnet2022yplg.png',dpi=500);
2 Japanese data example (The Ipython Notebooks are examples for playing with the models.)
Please first download the mseed data from
https://drive.google.com/drive/folders/1jAkW4kOvwUDYxXW-ty3BTY81fjLF7sKE?usp=share_link
Then download the manual picks from
https://drive.google.com/drive/folders/1KK16j1-WbwqKfTvh5gJbzJrX0fiayhl6?usp=share_link
Then run the notebook
notebooks/Test_Predict_MSEED_Japan.ipynb for the single-model (P and S together) example
notebooks/Test_Predict_MSEED_Japan_PS.ipynb for the double-model (P and S separately) example
The development team welcomes voluntary contributions from any open-source enthusiast.
If you want to make contribution to this project, feel free to contact the development team.
Regarding any questions, bugs, developments, collaborations, please contact
Yangkang Chen
[email protected]
The gallery figures of the eqcct package can be found at https://github.com/chenyk1990/gallery/tree/main/eqcct
Each figure in the gallery directory corresponds to a DEMO script in the "demo" directory. These gallery figures are also presented below.
DEMO1 The following figure shows an example of a comparison between different catalogs. Generated by demos/test_eqcct_texnet2022yplg.py
DEMO2 The following figure shows an example of a comparison between initial LOCSAT and refined NonLinLoc catalogs. Generated by demos/test_eqcct_locsatvsnll.py
DEMO3 The following figure shows an example of a comparison between different catalogs. Generated by demos/test_eqcct_nll_growclust.py
DEMO4 The following figure shows an example of a comparison between the one-month catalog before (up) and after (down) relocation, manual (left), and EQCCT (right). Generated by demos/test_eqcct_relo_manualvsAI.py
DEMO5 The following figure shows an example of comparison between one-year (manual, left) and one-month (EQCCT, right) catalog before (up) and after (down) relocation. Generated by demos/test_eqcct_relo_manualvsAI.py
DEMO6 The following figure shows an example of a comparison between the three-month (manual, left) and three-month (EQCCT, right) catalog after relocation. Generated by demos/test_eqcct_relo_manualvsAI.py
DEMO7 The following figure shows a test on station density's effect on earthquake detectability. Generated by demos/test_eqcct_stationeffect.py
DEMO8 The following figure shows plots of TexNet catalog of different periods (1 month, 3 months, 6 months, 1 year) on top of event density plot of the one-month EQCCT catalog. Generated by demos/test_eqcct_eventdistribution.py
DEMO9 The following figure shows a one-month continuous data test including station distribution, magnitude-frequency plot, depth distribution, daily detection distribution between the one-month TexNet, reported (detected via an optimized STA/LTA algorithm without manual quality control), EQCCT catalogs. Generated by demos/test_eqcct_eventdistribution.py
