Scripts to reproduce the results of "The precursory phase of large earthquakes" (Science, 2023)

This repository contains the scripts to reproduce the results of Bletery and Nocquet (Science, 2023).

Directories had to be zipped to be downloaded. Please start by unzipping them.
Complementary figures to the one published can be found (without running any script) in the following directories: eq_stack_figures (individual stacks of the 90 earthquakes), eq_stack_figures_excluded (stacks of the 4 earthquakes excluded from the study because of obvious offsets in the 2-day period preceding the events), test_figures (figures of the random tests for which r>1.82), figures_tohoku_test (figures of random tests for which the RMS reduction of a sinusoidal fit > 60%).

If you wish to run the scripts and reproduce the results, you will first need to install the pyacs and pyeq libraries. All the instructions can be found on the following GitHub repository: https://github.com/JMNocquet/pyacs36.

Then run the following scripts (you can skip the heavy downloading and subsequent steps by unzipping the station_location.zip file as well as the data_byEQ folder):

./make_station_list.py # Take a list of earthquakes (downloaded and reformated from the SCARDEC database: http://scardec.projects.sismo.ipgp.fr/). The list used in the article is composed of all earthquakes with depth < 60 km from 01 01 2000 to 12 31 2020 (time period covered by the catalog at the time of the work).

sort download_data.sh | uniq > download_data_uniq.sh # Suppress duplicate requests to avoid downloading the same files multiple times

chmod +x download_data_uniq.sh

./download_data_uniq.sh # Download UNR data (the total size of the data to download is > 20 GB, this will take some time)

./unzip_UNR_data.sh # Unzip annual files into daily files

./extract_UNR_time_series.py # Extract 48 h time series before the earthquakes

./make_synthetics.py # Calculate the synthetic displacements expected from hypothetical precursory slip

./make_dot_product.py # Calculate the dot products and stack them by earthquake

Plot the results in the notebook file:

jupyter notebook make_stack.ipynb # Calculate global stack and generates Figures 2-3, S1-S6

./make_files_4_map_plot.py # Generate files to plot Figure 1

./mapplot.gmt # Generate Figure 1 (needs gmt installed)

Make the test to estimate how frequently the signal we observe could randomly arise from noise:

./make_test_1.py # Calculate dot product stack for every earthquake for 4 "fake earthquakes" per day during 1 year

./make_test_2.py # Draw random combinations of "fake earthquakes" to estimate how frequently the signal we observe could randomly arise from noise

Make the test to estimate how frequently the signal we observe before Tohoku could randomly arise from noise (in notebook):

jupyter notebook tohoku_test.ipynb