Skip to content

These files are used to perform a huddle test with seismometers. It's a relative calibration of the poles, zeroes and the Gain of the sensors.

License

Notifications You must be signed in to change notification settings

marzeck/seismo-huddle

Repository files navigation

seismo-huddle

These files are used to perform a huddle test with seismometers. It's a relative calibration of the poles, zeroes and the Gain of the sensors.

Requirements

Background

Input are continuous seismic data of two or more seismometers in very close distance (cm) to each other measuring hours of the same noise. Best results are achieved with a big earthquake during sensing period. Only compare data of the same component! The example data provided in the repository is from a huddle test in Potsdam, Germany. Only vertical components are used around the time of a M6.7 earthquake in Brazil. In case you want to use a different dataset and/or a different event, please change in the file "huddle-test-T2.py" if necessary:

  • datadir
  • start(time) & end(time)
  • all Poles and Zeros (PAZ) and sampling frequency

workflow

1.)

$ python huddle-test-T2.py

Calculates all relative transferfunctions of all seismometer combination of the huddle test. The estimated transfer function (based on Havskov Instrumentation in Earthquake Seismology', p. 293-294) will be stored in a .txt-file in a separate subfolder (here /transfer-functions). Adjust input and output names in file.

2.)

$ bash huddle-dinver-run.sh

(uses script "huddle-test-optimizer.py")

A forward modeling of all given transferfunctions will be done with dinver. First, the python script "huddle-test-optimizer.py" will be adjusted automatically. Second, dinver will use this python script to optimize the Poles and Zeros and Gain. All free parameters, which will be optimized are provided in the .dinver-file and can be changed in there (with the use of dinver). The communication between the python script to be optimized and the forward modelling of dinver is done via the files "parameters" and "misfit". They are automatically created/removed and should not be changed during the calculations. The results (poles, zeroes, Gain, misfit) of each calculation step are stored in log files in the subfolder /misfits-log. An overview of the forward models can be found in run-reports.

New in Geopsy 3.4 and newer!! Dinver now allows multi-threading, by creating random (hexadecimal labelled) subdirectories and calls the external script in there. For that, absolute pathnames need to be provided to properly link between the external script (huddle-test-optimizer.py) and dinver. The 'parameter' and 'misfit' files are not affected and should remain with relative names. But you need to adapt the location of the external script within the huddle-test.dinver environment file, either with the GUI of dinver or convert the *.dinver file into an xml file and back:

$ tar -xvzf huddle-test.dinver
$ tar -cvzf huddle-test.dinver contents.xml 

3.)

$ python sort-misfits.py

The log files of the forward modelling are sorted in ascending order by their misfits.

4.)

$ bash huddle-results.sh

The uppermost lines (the results of smallest misfit) of all given ordered log-files are stored separately in the file "huddle-results.txt"

5.)

$ python huddle-analysis.py

A small python script to plot all poles (imaginary over real part) derived from forward modelling, with smallest misfit (see results section).

Results

resulting plot The resulting plot, as shown above, will give show all poles in different subplots. Since this example was performed with Lennartz instruments only the three poles are shown (zeros are three complex zeros). In the upper left panel, the whole complex plane is shown with all poles. The other three panels show each pole in close up for visual control.

In the huddle-analysis.py file the theoretical poles and zeros can be adjusted at the top of the file (in case you don't know the instrument specifications...). Those theoretical values will be shown as green diamonds in the plot. In addition, one station/sensor can be chosen to be highlighted (variable highl) in a different color.

This workflow inverted for poles, zeros and the gain, but the latter is not visually controlled yet. Feel free to suggest something for that. The values can be found in the huddle-result.txt file in the second last column.

Citation

Havskov, J., & Alguacil, G. (2004). Instrumentation in earthquake seismology.

About

These files are used to perform a huddle test with seismometers. It's a relative calibration of the poles, zeroes and the Gain of the sensors.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published