Various array processing tools for infrasound and seismic data. By default uses
least-squares to determine the trace velocity and back-azimuth of a plane wave
crossing an array in sliding time windows. More advanced processing (such as
least-trimmed squares) is easily integrated. Also provides tools to characterize
the array response, uncertainty, source-location of a spherical wave crossing
the array, etc. See
documentation and
example.py
for more info.
General References and Suggested Citations
Least squares and array uncertainty:
Szuberla, C. A. L., & Olson, J. V. (2004). Uncertainties associated with parameter estimation in atmospheric infrasound arrays, J. Acoust. Soc. Am., 115(1), 253–258. https://doi.org/doi:10.1121/1.1635407
Least-trimmed squares:
Bishop, J. W., Fee, D., & Szuberla, C. A. L. (2020). Improved infrasound array processing with robust estimators, Geophys. J. Int., 221 p. 2058-2074. https://doi.org/10.1093/gji/ggaa110
We recommend you install this package into a new
conda environment.
(Please install Anaconda or
Miniconda before proceeding.)
The environment must contain all of the packages listed in the
Dependencies section. For ease of installation, we've provided
an
environment.yml
file which specifies all of these dependencies as well as instructions for
installing array_processing itself. To install array_processing in this
manner, execute the following commands:
git clone https://github.com/uafgeotools/array_processing.git
cd array_processing
conda env create -f environment.yml
This creates a new conda environment named uafinfra and installs
array_processing and all of its dependencies there.
The final line in the environment.yml file installs array_processing in "editable" mode, which
means that you can update it with a simple git pull in your local repository.
We recommend you do this often, since this code is still under rapid
development.
For installation into a pre-existing conda environment, click here.
First ensure you have ObsPy and FastKML installed (
conda install -c conda-forge
obspy fastkml) and then download and install the uafgeotools
dependencies and this package with:
pip install git+https://github.com/uafgeotools/waveform_collection.git
pip install git+https://github.com/uafgeotools/lts_array.git
pip install git+https://github.com/uafgeotools/array_processing.git
(Note that this option does not produce a local clone of the repository.)
uafgeotools packages:
Python packages:
Import the package like any other Python package, ensuring the correct environment is active. For example,
$ conda activate uafinfra
$ python
>>> import array_processing
Documentation is available online
here. For a usage example, see
example.py.
Note: The array_plot() function does not allow both sigma_tau and dropped elements from least trimmed squares to be plotted.
The sigma_tau variable is an indicator of nonplanar propagation across an array (using all elements), and least trimmed squares drops element pairs that appear "too far" from planar. In this way, having a large sigma_tau value and having consistently dropped element pairs (while not the same) suggest a departure from the plane wave model.
sigma_tau is only calculated when ordinary least squares (ALPHA=1.0) is used. The ability to plot one or the other was intended as a safeguard against potentially conflicting processing assumptions. To maintain a consistent output data structure, the sigma_tau key returns a np.nan in the case that subset pairs are trimmed (0.5 <= ALPHA < 1.0).
If ALPHA=1.0, the dropped stations are not plotted since least trimmed squares is not used, and sigma_tau may be plotted if specified. If ALPHA < 1.0, then sigma_tau is not plotted or calculated.
(Alphabetical order by last name.)
Jordan Bishop
David Fee
Curt Szuberla
Liam Toney
Andrew Winkelman