mecompute

Program to compute Magnitude Energy (Me) from downloaded seismic waveforms:

• It downloads data (waveform segments) and metadata from the FDSN GEOFON event web service (using stream2segment)
• It computes the Me for each segment producing a tabular data (one row per segment) stored in HDF format (using stream2segment)
• It produces even-based HTML reports from each HDF table, with Me information and station Me residuals visualized on an interactive Map

Installation:

Make virtualenv python -m venv [PYPATH].

Clone three repositories: this one, stream2segment (for downloading and processing data) and sdaas (for computing the waveform anomaly score and discard outliers). We suggest to create a root directory and put all three therein

Then install the requirements. From the directory where you cloned mecompute:

    grep ./requirements.txt numpy
    pip install <what numpy is there>
    pip install -r ./requirements.txt
    (move to stream2segment repo)
    pip install -e .
    (move to sdaas repo)
    pip install -e .
    (move to this project repo)
    pip install -e .

Usage:

With your database created and, if needed, configured (you can use Postgres or SQLite), copy download.yaml into download.private.yaml (the file is ignored by git) and set the db url safely (the url might contain passwords).

Download:

[PYPATH]/bin/python [S2SPATH]/stream2segment/cli.py download -c [MEPATH]/s2s_config/download.private.yaml

where [S2SPATH] the path of the stream2segment repository (where you cloned stream2segment) and [MEPATH] the path of this project, then:

Process

[PYPATH]/bin/python [MEPATH]/cli.py process [ROOT]

where [ROOT] can be any directory of your choice.

This command processes by default the data downloaded in the previous day and creates under [ROOT] a process directory mecomputed--[START]--[END] (where [START] and [END] are the ISO-formatted time bounds of the previous day) with the following files:

• mecomputed--[START]--[END]
  • process--[START]--[END].hdf
  • process--[START]--[END].yaml
  • process--[START]--[END].log

(If you want to customize the time bounds, provide the relative options -s and -e. Type [PYPATH]/bin/python [MEPATH]/cli.py process --help for details)

Report

[PYPATH]/bin/python [MEPATH]/cli.py report [ROOT]

This command will scan each process directory under [ROOT] and create a .html report file next to each .hdf found.

[ROOT] is must just be the same directory given in the process routine (see details above).

The program does not overwrite existing HTML unless the -f option is given (type [PYPATH]/bin/python [MEPATH]/cli.py report --help for details).

You can pass as [ROOT] also a specific HDF file in case you want to regenerate a single report.

Misc

Generate test HTML report (to inspect visually):

Run test_workflow::test_report_fromfile and inspect test/data/process.result.multievent.html test/data/process.result.singleevent.html

Change the event URLs (for developers only)

This is not a foreseen change in the short run but better keep track of it to save a lot of time in case.

For the download and process part, where the program delegates stream2segment, you can change the event web service by simply changing the parameter eventws in the download.private.yaml file with any valid FDSN event URL.

The problem is the HTML report: currently, we hard code in the Jinja template (report.template.html) two URLs, related but not equal to eventws:

1. In each table row, an URL redirects to the event source page
2. In the map, an URL is queried to get the Moment tensor beach ball (which is used as event icon on the map)

Ideally, one should remove the hard coded URLs and implement Python-side a class that, given the eventws URL in download.private.yaml and an event_id, returns the two URLs 1) and 2) above, considering the case that any of those URLs might not exist, and thus think about fallbacks for the missing anchor in the table and the missing icon in the map