SkillAgentSearch skills...

Microseismic

A machine learning model for identifying rockfall signals from microseismic data using waveform guidance.

Install / Use

/learn @yuxi-chenc/Microseismic
About this skill

Quality Score

0/100

Supported Platforms

Universal

README

Intelligent Recognition Network for Microseismic Signals based on Waveform Attributes (SSD)

This repository provides a complete deep learning pipeline for seismic data processing, including label generation, model training, and event prediction, as presented in our corresponding research paper.

Features

  • Label Generation: Scripts to process raw seismic data and generate various types of labels (e.g., event-based and attribute-based)
  • Flexible Model Training: Training script supporting building models from scratch or fine-tuning pre-trained weights
  • Universal Event Prediction: Prediction script to apply trained models on new, continuous seismic data
  • Modular Workflow: Entire pipeline controlled through central runner script (run.py)

Requirements

To run this project, first install necessary Python libraries (recommended in a virtual environment):

pip install tensorflow numpy obspy pandas scipy matplotlib pywavelets

How to Run

All functionalities are integrated into run.py. Modify the run_case variable at the top of the script:

run_case = "train"  
Options for run_case:
eve      : Generates event labels (Noise, Event Type) from raw seismic data
att      : Generates attribute labels (e.g., SSD) from waveform data
pre_train: Pre-trains a model on attribute labels
train    : Trains/fine-tunes main model on event labels
pred     : Makes predictions on new data

Execution Steps

  • Choose Task: Set run_case in run.py to desired task (e.g., run_case = "pred")
  • Configure Parameters: Modify command-line parameters in corresponding if/elif block
  • Run Script:
python run.py

Model

-Architectures: Defined in models.py and models_f.py -Pre-trained Weights: Provided in model/ directory

Dataset

All data and code are openly available.

  • Curated datasets from paper (download and place in appropriate directory): Download Link: https://doi.org/YOUR_DOI_HERE

-Continuous Raw Data Sources Séchilienne Rockslide OMIV/RESIF Use day_datadown.py script https://doi.com/10.15778/RESIF.MT Illgraben Rockslide GFZ Data Services Manual download https://doi.com/10.5880/GFZ.2.4/2016.001.

Reference:

Bianchi, M., Evans, P. L., Heinloo, A., & Quinteros, J. (2015). Webdc3 web interface. GFZ Data Services. doi: 10.5880/GFZ.2.4/2016.001388 Helmstetter, A., & Garambois, S. (2010). Seismic monitoring of S´echilienne rockslide (French Alps): Analysis of seismic signals and their correlation with rainfalls. Journal of Geophysical Research: Earth Surface, 115 (F3). doi: 10.1029/2009JF001532409 French Landslide Observatory – Seismological Datacenter / RESIF. (2006). Observatoire Multi-disciplinaire des Instabilit´es de Versants (OMIV) [Data set]. RESIF - R´eseau French Landslide Observatory – Seismological Datacenter / RESIF. (2006). Observatoire Multi-disciplinaire des Instabilit´es de Versants (OMIV) [Data set]. RESIF - R´eseau

Related Skills

View on GitHub
GitHub Stars5
CategoryEducation
Updated11d ago
Forks0

Languages

Python

Security Score

85/100

Audited on Mar 23, 2026

No findings