/FocalMechClassifier

Codes for classifying to focal mechanisms of earthquakes.

Primary LanguagePythonMIT LicenseMIT

FocalMechClassifier

For reproducibility, we demo the code using the dataset provide with the USGS HASH code, http://earthquake.usgs.gov/research/software/index.php.

Download the hash.v1.2.tar.gz, unpack it and copy the north1.phase and scsn.reverse datafiles into the "demo" directory of this project. This will be the inputs to the demo test[2-6].py, run these from the demo directory.

You'll also need to compile the hash_driver1 exe is you want to run it. You can use this to regenerate the example1.inp, example1-norev.inp data. Do,

cat test1.out | sed -e s/*//" > example1.out

this removes the extra column that messes up genfromtxt.

We provide the input file example1.inp, example1-norev.inp input parameters. And also data derived from them, 3146815_realizations.out, example1.out and example1-noreverse.out. You can regenerate these using the inputs provided.

If you get "LookupError: unknown encoding: " errors from ObsPy, set these environment variables:

export LC_ALL=en_US.UTF-8

export LANG=en_US.UTF-8

This seemed to effect OSX: El Capitan.