This repository provides reproducible materials for Seismic reservoir characterization with implicit neural representations by authors Romero J., Heidrich W., Luiken N., and Ravasi M.
The repository is organized as follows:
- 📂 intraseismic: A Python library that includes routines for dataset management, different types of coordinates encoding, the IntraSeismic model, train functions, and plotting functions.
- 📂 data: A folder containing the data or instructions on how to obtain it.
- 📂 notebooks: Jupyter notebooks that document the application of IntraSesimic to the inversion of the synthetic Marmousi data.
The provided notebooks include:
- 📂 Marmousi
- 📙
Marm_data_creation.ipynb: Creates post-stack synthetic seismic datasets with varying noise levels for the Marmousi model. - 📙
Poststack_IS_Marm.ipynb: Demonstrates the inversion of Marmousi seismic data with a noise level of$\sigma = 0.1$ using IntraSeismic. - 📙
Poststack_IS_Marm_MCUQ.ipynb: Conducts Monte-Carlo Dropout uncertainty quantification in IntraSeismic. - 📙
Prestack_IS_3nets_Marm.ipynb: Pre-stack seismic inversion of Marmousi model using IntraSeismic.
- 📙
To reproduce the results, use the environment.yml file for environment setup.
Execute the following command:
./install_env.sh
The installation takes some time. If you see Done! in your terminal, the setup is complete.
Finally, run:
pip install -e .
in the folder where the setup.py file is located.
Always activate the environment with:
conda activate my_env
Disclaimer: Experiments were conducted on an AMD EPYC 7713 64-Core processor equipped with a single NVIDIA TESLA A100. Different hardware may require alternate environment configurations.