Ridgecrest sequence FakeQuakes synthetic displacement waveforms and real GNSS data

Datasets produced and used for the paper Detecting Earthquakes in Noisy Real-Time GNSS Data with Deep Learning for Improved PGD Magnitude Estimation (2026) by Sydney N. Dybing, Diego Melgar, Amanda M. Thomas, Dara E. Goldberg, David Mencin, and Brendan W. Crowell submitted to Bulletin of the Seismological Society of America. Codes used to produce and process these data are available at https://github.com/UO-Geophysics/gnss-picker. Files in this repository: newfault_fq_wvfm_data_formatted.tar.gz (hdf5 format): FakeQuakes waveforms produced from the Ridgecrest fault structure used in training and testing newfault_fq_wvfm_info.tar.gz (npy format): Metadata associated with the FakeQuakes waveforms. [FQ rupture name, station name, earthquake magnitude] summer23_128samps_all_noise_samples.tar.gz (hdf5 format): Noise samples selected from the real Ridgecrest sequence GNSS data demean_realdata_rembad.tar.gz (hdf5 format): Demeaned real Ridgecrest sequence GNSS data real_metadata_rembad_w_gauss_pos_mag.tar.gz (npy format): Metadata associated with the real Ridgecrest sequence GNSS data. [station name, date sample came from, sample start time, sample end time, random n counter, P-wave arrival index, earthquake magnitude]. Samples that are just noise (not earthquakes) have 'nan' for the last two columns. newfault_gfast_test_alltestdata.tar.gz (hdf5 format): FakeQuakes waveforms reformatted for the GFAST test described in Dybing et al. (2026). newfault_gfast_test_alltestdata_info.tar.gz (npy format): Metadata for the FakeQuakes waveforms reformatted for the GFAST test. [FQ rupture name, station name, earthquake magnitude, P-wave arrival index] bestmodel_traindate_2024-10-01.tar.gz (h5 format): Best model trained in the Dybing et al. (2026) study