Mparloc: Bayesian Method for Earthquake Location Using Multi-Parameter Data

(c) 2020-2021 Alessandro Caruso alessandro.caruso@unina.it,

Post-doctoral researcher at RISSCLab - Department of Physics (University of Naples Federico II)

DOI

Introduction

I implement a Bayesian method for the earthquake location, which combines the information derived from the differential P-wave and S-wave arrival times, amplitude ratios and back-azimuths measured at a minimum of two stations. A complete posterior pdf of location is provided as a probability map and uncertainty of location is estimate as the range where the probability is over a prefixed threshold. In order to improve the speed performance, a pseudo-global search algorithm is implemented to explore in a fast way an even very dense computational grids.

Config, Installing and Running

Mparloc is written in pure python3 code tested with version 3.7.3 and several external library dependency installed:

obspy, numpy, nllgrid, math, pyproj, taup, matplotlib

Mparloc library consists of two main functions implemented in the source file Mparloc.py :

grid_creatore(Name_network,Name_event) # Tool for creating grids starting from NnLinLoc grids.
Launcher(Name_network,Name_event) # Function that initialize and launch the location algorithm.

Name_network and Name_event are string objects that contain respectively the name of the network folder and the name of records folder. The network folder contains: a sub-folder grids, a records folder and a text file named list_stations_4loc.dat that is the station name list used by Mparloc. The grids folder contains .buf and .hdr files of time grids created with tools of Nonlinloc, while in the records folder there are the accelerometric or velocimetric sac files with their station fields and transduction coefficient correctly fixed in header. Following an example of list_stations_4loc.dat:

AND3
AVG3
BEL3
BENI
BSC3
CGG3 ....

The configuration of module is done by compiling two files located in the config_files folder:

run_loc.conf # Main configuration file
weights_table.dat # Correspondence table between SNR(Signal to Noise Ratio) and the error in second on seismic pick phase

Example of weights_table file:

# SNRmin SNRmax Wp(error on P) Ws(error on S)
10 0.25 0.55  # In the first line there is only SNRmax
10 20 0.15 0.25
20 50 0.1 0.15
50 0.015 0.015 # In the last line there is only SNRmin

Example of run_loc file:

#Setting parameters:

#Parameter to set grid computation precision [f for single-precision, d for double-precision]
grid_precision: f
#Confidence level for error estimate [float number between 0 and 1]
Confidence_probability_threshold_level: 0.5
#Time window in seconds for SNR estimate for P phase [float positive number]
snr_wind_p: 0.5
#Time window in seconds for SNR estimate for S phase [float positive number]
snr_wind_s: 0.5
#Minimum number of P phases for eqk location [integer positive number >2]
no_min_p: 2
#Minimum number of S phases for eqk location [integer positive number >1]
no_min_s: 1
#Minimum SNR to reject low quality data [integer positive number]
min_SNR: 0
#Use back-azimuth data for eqk location [False or True]
back_az: True
#back-azimuth standard deviation in degree [integer positive number]
back_az_error: 30
#Use Pv differential amplitude data for eqk location [False or True]
diff_Amp: True
#Phase Amplitude attenuation coefficient [float negative number]
diff_Amp_b: -1.78
#differential amplitude standard deviation [float positive number]
diff_Amp_error: 0.5
#Use S phases for eqk location [False or True]
s_phases: True

After configuring the software and correctly formatting input files, it is possible to execute the tool to perform the building of grids. For example, naming the network folder as ISNet and the event folder as 12458a:

#!/usr/bin/python3
#IMPORT OF LIBRERY
from MParLoc import grid_creator
# RUN THE GRID CREATOR CODE THAT READ THE NONLINLOC GRIDS 
#AND CREATE DIFFERENTIAL AMPLITUDE AND BACK-AZIMUTH GRIDS
grid_creator('ISNet','12458a')

Finally you can run the location algorithm:

#!/usr/bin/python3
#IMPORT OF LIBRERY
from MParLoc import Launcher
# RUN THE LOCATION ALGORITHM
Launcher('ISNet','12458a')

After running, the code provide in the folder of event 12458a an hypo71 format file MParLoc_event.h71 with all available information about the event location, the location Map MParLoc_event_FIGURE.pdf which gives the shape of the probability spatial trend and a textual log MParLoc_event.log of main operations performed by the algorithm.

Testing

After the installation of all dependence, you can test the library run the follow command in the main folder:

python3 Example.py (Ubuntu environment)
python Example.py (Windows environment)

if the code work well the output files are provided at path: @localpath/Mparloc/EXAMPLE/sac/

Important Note

This is a research implementation of the code, with no actual real-time capabilities. Feel free to pick the relevant parts of this (GPL licensed) code and build your own real-time module!

References

Zollo, A. , Caruso, A. , De Landro , G. , Colombelli, S. , Elia, L. 2020. A Bayesian Method for Real-time Earthquake Location Using Multi-Parameter Data. JOURNAL OF GEOPHYSICAL RESEARCH: SOLID EARTH doi: [https://doi.org/10.1029/2020JB020359]

DISCLAIMER

THIS SOFTWARE IS PROVIDED "AS IS" WITHOUT WARRANTY.