Related paper:

Zhu, W., Mousavi, S. M., & Beroza, G. C. (2018). Seismic Signal Denoising and Decomposition Using Deep Neural Networks. arXiv preprint arXiv:1811.02695.

1. Install

The code is tested under python3.6

Using virtualenv

pip install virtualenv
virtualenv .venv
source .venv/bin/activate
pip install -r requirements.txt

Using ananconda

conda create --name venv python=3.6
conda activate venv
conda install tensorflow=1.10 matplotlib scipy pandas tqdm

2. Demo data

Located in directory: Dataset/

3. Model

Located in directory: Model/190614-104802

4. Prediction

Data format:

Required a csv file and a directory of npz files.

The csv file contains one column: "fname"

The npz file contains one variable: "data"

The variable "data" is one component sample with a length of 3000 with sampling rate of 100 Hz.

python run.py --mode=pred --model_dir=Model/190614-104802 --data_dir=./Dataset/pred --data_list=./Dataset/pred.csv --output_dir=./output --plot_figure --save_result --batch_size=20

Notes:

1. For large dataset and GPUs, larger batch size can accelerate the prediction.
2. Plotting figures is slow. Removing the argument of --plot_figure can speed the prediction

5. Train

Data format

Required: two csv files for signal and noise, corresponding directories of the npz files.

The csv file contains four columns: "fname", "itp", "channels"

The npz file contains four variable: "data", "itp", "channels"

The shape of "data" variables has a shape of 9001 x 3

The variables "itp" is the data points of first P arrival times.

Note: In the demo data, for simplicity we use the waveform before itp as noise samples, so the train_noise_list is same as train_signal_list here.

python run.py --mode=train --train_signal_dir=./Dataset/train --train_signal_list=./Dataset/train.csv --train_noise_dir=./Dataset/train --train_noise_list=./Dataset/train.csv --batch_size=20

Please let us know of any bugs found in the code. Suggestions and collaborations are welcomed!