Synthetic dataset of 2342 earthquake/tsunami scenarios targeting the Nankai trough subduction zone

Summary:

• Data of 2342 hypothetical Nankai Trough events
• Tsunami propagation is simulated by the GeoClaw software (Clawpack Development Team, 2021).
• Earthquake realizations generated by the Mudpy software (Melgar, 2020).
• Data composition is as follows:
  nankai_data
├── quakes
│   ├── dtopofiles
│   │   ├── nankai_xy_zzzzzz.dtt3
│   │   ├── ...
│   ├── png
│   │   ├── nankai_xy_zzzzzz_slip_dtopo.png
│   │   ├── ...
│   └── ruptures
│       ├── nankai_xy_zzzzzz.log
│       ├── nankai_xy_zzzzzz.rupt
│       ├── ...
├── waves
│   ├── nankai_xy_zzzzzz.csv 
│   ├── ...
├── gauge_loc.csv
└── wave_seq.npy

Details of each file:

Unzipping `nankai_data.tar.gz` creates two directories and two file: `quakes/`, `waves/`, `gauge_loc.csv` and `wave_seq.npy`.
Note that 8 GB of additional storage is required. 

• quakes/

  Earthquake data of 2342 scenarios. 
  • quakes/dtopofiles/nankai_xy_zzzzzz.dtt3
    Seafloor deformation of each scenario
    Mw: x.y, ID: zzzzzz
  • quakes/ruptures/
    nankai_xy_zzzzzz.rupt stores the rupture for each scenario.
    nankai_xy_zzzzzz.log has information about the rupture.
  • quakes/png/nankai_xy_zzzzzz_slip_dtopo.png
    The slip distribution and seafloor deformation due to fault rupture are visualized in PNG format.
    
    

• waves/

  2342 .csv files contain time series data of simulated tsunami wave heights
  The left column has the time (minute) elapsed from the fault rupture
  The following columns have the wave sequences recorded at each gauge
  Each file has 2160 rows corresponding to the simulation time steps, 3 [hr] x 3600 [sec/hr] / 5[sec] = 2160.
  
  

• gauge_loc_all.csv

  The locations of 62 synthetic gauges
  Gauge IDs are in the left column named ID.
  The gauge locations are in Longitude and Latitude columns.
  Some of the synthetic gauges are set by referring to the locations of existing gauges, as shown in the Instruments column.
  
  

• wave_seq.npy

   Data matrix containing the wave sequences. 
   Loading this binary data by 
       numpy.load('wave_seq.npy') 
   gives a 2342 x 62 x 2160-shaped 3d array.