1. Introduction to the EQcorrscan package¶
This document is designed to give you an overview of the capabilities and implementation of the EQcorrscan Python package.
1.1. Why EQcorrscan?¶
EQcorrscan is designed to compute detections of earthquakes, or any seismic signal (explosions work really well) using more advanced routines than standard amplitude-ratio methods.
This package was originally based around a matched-filter detection routine which works by comparing templates with continuous data. The main benefit of EQcorrscan’s matched-filter routine is the level of parallel processing that can be achieved. EQcorrscan will run on anything from a 1GB RAM single-board computer to a multi-hundred-GB RAM, thousand CPU high-performance computer. Because the internals of EQcorrscan’s matched-filter routine scale reasonably well, the developers have observed speed-ups of 150x (from 2 months to 10 hours) by migrating from a small cluster to a large one (for a 6.5 year long continuous dataset and 800 templates).
The authors of EQcorrscan foresee this project as an open repository for the development of software for the detection and analysis of repeating and near-repeating earthquakes. This repository will continue to grow and develop and any and all help/criticism will be appreciated.
There are a lot of things that could be added to this project - if you want to get involved the best place to start, and the most valuable thing for your understanding, and for the health of this package would be to contribute tests and documentation.
1.2. Installation - Updated for version 0.2.x¶
In general we recommend users to install EQcorrscan in a virtual environment, conda will simplify your install greatly - we recommend creating a conda environment with the following:
conda create -n eqcorrscan colorama numpy scipy matplotlib obspy bottleneck pyproj source activate eqcorrscan
1.2.1. Non-Python dependancies: Ubuntu:¶
Prior to installing the python routines you will need to install the fftw library. On linux use apt (or your default package manager - note you may need sudo access):
apt-get install libfftw3-dev
1.2.2. Non-Python dependancies: OSX:¶
For OS-X systems install fftw from conda, available via the menpo channel:
conda install -c menpo fftw
On OSX you will have to ensure that the gcc used for install is an actual gcc, not clang. To do this you should also install gcc into your conda environment:
conda install gcc
Note that you can install fftw and gcc from other sources, however, we know there is an issue with homebrew gcc-4.9 (but not with macports gcc-4.9) - other gcc versions tested do work however.
1.2.3. Non-Python dependancies: Windows:¶
For Windows systems you should follow the instructions on the fftw-windows install page and use the pre-compiled dynamic libraries. These should be installed somewhere on your system path, or the install location added to your path. The correlation routines use openMP for parallel workflows, however, some aspects of this run into issues with version of MSVC < 10.0 (due to old C standards being used), as such, by default, the correlation routines are compiled as serial workflows on windows. If you have a need for this threading in windows please get in touch with the developers.
1.2.4. Final EQcorrscan install:¶
Once you have installed fftw the EQcorrscan install should be as simple as:
pip install eqcorrscan
Note you may have issues with these installs if you don’t have numpy installed: but if you don’t have numpy installed then you have bigger issues...
If you plan to run the bright_lights or generating a synthetic grid of templates you will need to have grid csv files, which the authors have previously used NonLinLoc to generate. This is not provided here and should be sourced from NonLinLoc. This will provide the Grid2Time routine which is required to set-up a lag-time grid for your velocity model. You should read the NonLinLoc documentation for more information regarding how this process works and the input files you are required to give.
1.3. Supported environments¶
We support Linux, OSX and Windows environments running Python 2.7, 3.4 and 3.5. We don’t run our tests on other versions of Python so you might have some issues with other Python 3.x series, if you do, let us know.
We do not support Python 2.6.
This package is divided into sub-directories of core and utils. The utils directory contains simple functions for integration with Seisan, these are in the sfile_util module and functions therein which are essentially barebones and do not have the full functionality that seisan can handle. utils also contains a simple peak-finding algorithm findpeaks which looks for peaks within noisy data above a certain threshold and within windows.
Many other functions have been added to this module to handle the analysis of repeating and near-repeating earthquakes, including stacking routines, clustering algorithms, magnitude calculation both by amplitude picking and by singular value decomposition. I recommend you take a look in here to see if any of it is useful. There are also some plotting routines that make handling large datasets a little simpler. Most recently I have added a simple synthetic seismogram generator, which is currently my main project focus.
Since earlier versions the core modules have moved away from using parameter files, and instead rely on explicit argument calls. The parameter files are still included by not documented here (see inside the par files), and remain useful when generating batch scripts (see the scripts in the github repo).
Within core you will find the core routines to generate templates, (template_gen) search for likely templates (bright_lights) and compute cross-channel correlations from these templates (match_filter). The bright_lights and match_filter submodules have been designed with parallel computing in mind, to the extent that the more cores and machines you have running them the better. These rely on the python multiprocessing module to handle parallelisation at lower-levels. You can also do some ‘brute-force’ parallelisation on a day level when computing detections over multiple days. I tend to run one day per node of a cluster computer, with each day running templates in parallel.
1.5. Running tests¶
One of the main goals of EQcorrscan is to improve reliability and reproducibility of earthquake detection. To this end, EQcorrscan has a moderate test-base (you can check how much of our codebase if tested by looked at the badges in the github repository). You can also run these tests yourself locally to ensure that everything runs as you would expect in your environment. Although every effort has been made to ensure these tests run smoothly on all supported environments (using the ci bots), if you do find any issues, please let us know on the github page.
To run the tests you will need to have pytest installed along with a couple of extras (pytest-pep8 and pytest-cov). These can be installed by pip:
pip install pytest pytest-pep8 pytest-cov
You will also need to have a clone of the github repository:
git clone https://github.com/eqcorrscan/EQcorrscan.git
You can then run the tests from within the repository directory:
python setup.py test
Tests will take about half an hour to run (as of v.0.1.4) and will provide a coverage report at the end and notify you of any failures.