Dataset for "Reduced-order modeling for complex 3D seismic wave propagation"

This repository contains the seismic waveform data for the paper, "Reduced-order modeling for complex 3D seismic wave propagation", in Geophysical Journal International.

The dataset, velocity_time_series.npy, is a NumPy binary file and consists of 60-second, three-component, simulated velocity seismograms for 500 source locations and 8181 receivers. The simulations are for earthquakes in Southern California using a 3D velocity model, viscoelastic attenuation, and topography. The point source locations used for training are provided in the CSV file source_locations.csv. The SeisSol input files (mesh file, parameter file, material file, source files, and receiver file) necessary to produce the simulated data are provided in the zip file titled seissol_input_files.zip.

We also provide a smaller version of the dataset, seismos_16_receivers.npy, that only includes data for the 16 evenly spaced receivers that are identified in the paper. A Jupyter Notebook, seismogram_rom_demo.ipynb, is also provided that demonstrates how to create a reduced-order model that rapidly generates approximate seismograms.