SourceSpec - 1️⃣0️⃣0️⃣0️⃣ commits 🎂

Earthquake source parameters from P- or S-wave displacement spectra

Description

SourceSpec is a collection of command line tools to compute earthquake source parameters (seismic moment, corner frequency, radiated energy, source size, stress drop) from the inversion of P-wave and S-wave displacement spectra recorded at one or more seismic stations. SourceSpec also computes attenuation parameters (t-star, quality factor) and, as a bonus, local magnitude.

See Madariaga (2011) for a primer on earthquake source parameters and scaling laws.

Go to section Theoretical background below to get more information on how the code works. More details are available on the official SourceSpec documentation.

SourceSpec is written in Python and requires a working Python environment to run (see Installation below). However, since SourceSpec is based on command line, you don't have to know how to code in Python to use it.

The SourceSpec package is made of three command line tools:

source_spec: Compute earthquake source parameters from the inversion of P- or S-wave spectra.
source_model: Direct modelling of P- or S-wave spectra, based on user-defined earthquake source parameters.
source_residuals: Compute station residuals from source_spec output.

Getting Started

For the impatient

If you have seismic recordings in miniSEED format (e.g., traces.mseed), metadata in StationXML format (e.g., station.xml) and event information in QuakeML format (e.g., event.xml), then:

Generate a config file via source_spec -S;
Edit the config file variable station_metadata to point to station.xml file;
Run source_spec -t traces.mseed -q event.xml.

Note that the default config parameters are suited for a M<5 earthquake recorded within ~100 km. Adjust win_length, noise_pre_time, and the frequency bands (bp_freqmin_*, bp_freqmax_*, freq1_*, freq2_*) according to your setup.

Command line arguments

After successfully installed SourceSpec (see Installation below), you can get help on the command line arguments used by each code by typing from your terminal:

source_spec -h

(or source_model -h, or source_residuals -h).

source_spec and source_model require you to provide the path to seismic traces via the --trace_path command line argument (see File formats below).

Information on the seismic event can be stored in the trace header (SAC format), or provided through a QuakeML file (--qmlfile) or a HYPO71 or HYPOINVERSE-2000 file (--hypocenter). See File formats below for more information on the supported file formats.

Configuration file

source_spec and source_model require a configuration file. The default file name is source_spec.conf, other file names can be specified via the --configfile command line argument.

You can generate a sample configuration file through:

source_spec -S

Take your time to go through the generated configuration file (named source_spec.conf): the comments within the file will guide you on how to set up the different parameters.

File Formats

Trace formats

SourceSpec can read all the trace formats supported by ObsPy.

Two very common choices are:

miniSEED
SAC

The SAC format can carry additional information in its header, like event location and origin time, phase picks, instrument sensitivity.

Event formats

SourceSpec can read event information (event ID, location, origin time) in the following formats:

QuakeML: SourceSpec will also read phase picks and focal mechanism, if available
HYPO71
HYPOINVERSE-2000: SourceSpec will also read phase picks, if available

Event information can also be stored in the SAC file headers (header fields: EVLA, EVLO, EVDP, O, KEVNM).

Phase pick formats

Phase picks for P and S waves can be read from one of the following formats:

Phase picks can also be stored in the SAC file headers (header fields: A and T0).

Station metadata formats

Station metadata (coordinates, instrumental response) can be provided in one of the following formats:

Note that SEED RESP and PAZ formats do not contain station coordinates, which should therefore be in the trace header (traces in SAC format).

The station metadata file name or file directory is provided in the configuration file through the parameter station_metadata.

Alternatively, instrument sensitivity can be provided in the SAC header or as a constant in the configuration file. In both cases, use the configuration parameter sensitivity.

Output files

The SourceSpec main code, source_spec will produce the following output files (EVID is replaced by the actual event ID):

EVID.ssp.yaml: YAML file containing the estimated spectral parameters (summary values and per station values)
EVID.ssp.out (deprecated): text file containing the estimated spectral parameters (summary values and per station values)
EVID.ssp.log: log file in text format (including the command line arguments, for reproducibility)
EVID.ssp.conf: the input config file (for reproducibility)
EVID-residuals.pickle: station residuals in Python pickle format
EVID.ssp.h: hypocenter file in HYPO71 format with the estimated moment magnitude (only if an input HYPO71 file is provided)
EVID.xml: updated QuakeML file with the results of the SourceSpec inversion (only if an input QuakeML file is provided)

The following plots will be created, in png, pdf or svg format:

EVID.traces.png[.pdf,.svg]: trace plots
EVID.ssp.png[.pdf,.svg]: spectral plots
EVID.sspweight.png[.pdf,.svg]: spectral weight plots
EVID.boxplot.png[.pdf,.svg]: box plots for the earthquake source parameters retrieved at each station
Misfit plots, when using "grid search" or "importance sampling" for the spectral inversion

As an option, station maps can be created (requires Cartopy):

EVID.map_mag.png[.pdf,.svg]: station map with symbols colored by estimated moment magnitude
EVID.map_fc.png[.pdf,.svg]: station map with symbols colored by estimated corner frequency

As an option, the retrieved source parameters (per station and summary) can be appended to a SQLite database, whose path is defined in the configuration file.

Finally, always as an option, source_spec can generate a report in HTML format.

Theoretical Background

For each station, the code computes P- or S-wave displacement amplitude spectra for each component (e.g., Z, N, E), then combines the component spectra through the root sum of squares:

$$ S(f) = \sqrt{S^2_z(f) + S^2_n(f) + S^2_e(f)} $$

where $f$ is frequency and $S_x(f)$ is the P- or S-wave spectrum for component $x$.

It then inverts spectra for a 3-parameter Brune (1970) source model:

$$ S(f) = C \cdot \frac{M_0}{1+(f/f_c)^2} \cdot e^{-\pi f t^*} $$

where the three parameters to determine are:

the seismic moment $M_0$
the corner frequency $f_c$
the attenuation parameter $t^*$

and $C$ is a coefficient containing geometrical spreading, radiation pattern, seismic wave speed and medium density.

The inversion is performed in moment magnitude $M_w$ units (logarithmic amplitude). Different inversion algorithms can be used:

TNC: truncated Newton algorithm (with bounds)
LM: Levenberg-Marquardt algorithm (warning: Trust Region Reflective algorithm will be used instead if bounds are provided)
BH: basin-hopping algorithm
GS: grid search
IS: importance sampling of misfit grid, using k-d tree

Starting from the inverted parameters $M_0$ ( $M_w$ ), $fc$, $t^*$ and following the equations in Madariaga (2011), other quantities are computed for each station:

the Brune stress drop
the source radius
the quality factor $Q_0$ of P- or S-waves

Finally, the radiated energy $E_r$ can be measured from the displacement spectra, following the approach described in Lancieri et al. (2012).

As a bonus, local magnitude $M_l$ can be computed as well.

Event summaries (mean, weighted mean, percentiles) are computed from single station estimates. For mean and weighted mean estimation, outliers are rejected based on the interquartile range rule.

See the official documentation for more details.

Example three-component trace plot (in velocity), showing noise and S-wave windows

Example displacement spectrum for noise and S-wave, including inversion results

Installation

SourceSpec requires at least Python 3.6. All the required dependencies will be downloaded and installed during the setup process.

Using pip and PyPI (preferred method)

The latest release of SourceSpec is available on the Python Package Index.

You can install it easily through pip:

pip install sourcespec

From SourceSpec GitHub releases

Download the latest release from the releases page, in zip or tar.gz format, then:

pip install sourcespec-X.Y.zip

pip install sourcespec-X.Y.tar.gz

Where, X.Y is the version number (e.g., 1.2). You don't need to uncompress the release files yourself.

From SourceSpec GitHub repository

If you need a recent feature that is not in the latest release (see the unreleased section in CHANGELOG), you want to use the source code from the SourceSpec GitHub repository.

For that, clone the project:

git clone https://github.com/SeismicSource/sourcespec.git

(avoid using the "Download ZIP" option from the green "Code" button, since version number is lost), then install the code from within the sourcespec main directory by running:

pip install .

Documentation

The official SourceSpec documentation can be find at sourcespec.readthedocs.io.

Sample Runs

Several sample runs are available in the sourcespec_testruns repository.

Getting Help / Reporting Bugs

🙏 I need help

Join the SourceSpec Discussions and feel free to ask!

🐞 I found a bug

Please open an Issue.

Contributing

I'm very open to contributions: if you have new ideas, please open an Issue. Don't hesitate sending me pull requests with new features and/or bugfixes!

How to Cite

If you used SourceSpec for a scientific paper, please cite it as:

Satriano, C. (2022). SourceSpec – Earthquake source parameters from P- or S-wave displacement spectra (X.Y). doi: 10.5281/ZENODO.3688587

Please replace X.Y with the SourceSpec version number you used.

References

Brune, J. N. (1970). Tectonic stress and the spectra of seismic shear waves from earthquakes, J. Geophys. Res., 75 (26), 4997– 5009, doi: 10.1029/JB075i026p04997
Lancieri, M., Madariaga, R., Bonilla, F. (2012). Spectral scaling of the aftershocks of the Tocopilla 2007 earthquake in northern Chile, Geophys. J. Int., 189 (1), 469–480, doi: 10.1111/j.1365-246X.2011.05327.x
Madariaga, R. (2011). Earthquake Scaling Laws. In "Extreme Environmental Events", pp. 364–383, doi: 10.1007/978-1-4419-7695-6_22. Available on ResearchGate.

anton-br / sourcespec Goto Github PK

sourcespec's Introduction