This tool can be used to simulate noise cross-correlations and sensitivity kernels to noise sources.
Install requirements (easiest done with anaconda)
- obspy
- PyYaml
- pandas
- mpi4py
- geographiclib
- cartopy
- h5py
- jupyter
- pytest
- pyasdf
- psutil
Additionally, install instaseis, if you plan to use it for Green's functions (pip install instaseis). Currently it requires Python <= 3.8.
Install jupyter notebook if you intend to run the tutorial (see below).
If you encounter problems with mpi4py, try removing it and reinstalling it using pip (pip install mpi4py).
Clone the repository with git:
git clone https://github.com/jigel/noisi_inv.git
Change into the noisi/ directory. Call pip install . here, or call pip install -v -e . if you intend to modify the code.
After installation, change to the noisi/noisi directory and run pytest. If you encounter any errors (warnings are o.k.), we'll be grateful if you let us know.
To see an overview of the tool, type noisi --help.
A step-by-step tutorial for jupyter notebook can be found in the noisi/noisi directory.
Examples on how to set up an inversion and how to import a wavefield from axisem3d are found in the noisi/examples directory.
We have added a jupyter notebook to help you setup a config file for an inversion. All the different parameters are explained within that notebook. It is also available as python script.
It is recommended to download a pre-computed wavefield if you want to do proper inversions (http://ds.iris.edu/ds/products/syngine/). These are then easily implemented in the config file.