Archive of simulation data and results for Frontiers 2020 paper

This website collects data, results, and additional figures to accompany the paper

A Source Clustering Approach for Efficient Inundation Modeling and Regional Scale PTHA
by A.L. Williamson, D. Rim, L.M. Adams, R.J. LeVeque, D. Melgar, and F.I. González,
Frontiers in Earth Science, 8 (2020) p. 442.
10.3389/feart.2020.591663

Code archive: ptha_frontiers2020_code.tar.gz (1.5M)

Note that the 2000 hypothetical earthquakes were split into 4 magnitude classes:

Mw 7.5: Runs 000000 - 000499
Mw 8.0: Runs 000500 - 000999
Mw 8.5: Runs 001000 - 001499
Mw 9.0: Runs 001500 - 001999

These index numbers are used in organizing some of the results and plots linked below.

The GeoClaw results for each run are organized into the following...

Bathymetry/Topography images ... GeoClaw topofiles
Rupture images ... Seafloor deformation (dtopo) images ... GeoClaw dtopo files
Coarse-grid model run results: GeoClaw output directories ... Plots and animations
Fine-grid model run results: GeoClaw output directories ... Plots and animations

For the plots and animations, click on the link above and then navigate to ptha_00XXXX/_plots/_PlotIndex.html to find the plot index for the specified run.

Fakequakes .rupt files for the 2000 realizations can be found in ruptures.tar.gz.

Hazard curves and maps after clustering

Comparison 4 contains the paper results. In those images, the strategy names strategy-all-fine, strategy-all-coarse-modI, strategy-C5, and strategy-all-pseudoI-C5 correspond to the names all-fine, coarse-mod, clusters-fine, pseudo-fine, respectively, that were used in the paper.

Other comparsions were based on different clustering strategies and/or compare different approaches to approximating the hazard curves.

See also the transect comparisons for C5, some of which appeared in the paper. C5 refers to the clustering strategy used in the paper. Transects are also included that show using 18 clusters is an improvement over the 6 clusters of C1.

These comparisons led to our adoption of the C5 clustering with 18 clusters for the paper.

Comparison 1: Cluster C1, Comparing to coarse grid alone, and to coarse-mod results: Westport Crescent City
Comparison 2: Cluster C1, Comparing to coarse-mod and pseudo-fine results Westport Crescent City
Comparison 4: Cluster C5, Comparing to coarse-mod and pseudo-fine results Westport Crescent City
Comparison 5: Cluster C5, Comparing to pseudo-fine results Westport Crescent City
Comparison 20: Comparing C5 and C6 cluster representative strategies Westport Crescent City
Comparison 21: Comparing C1new, C3, and C5 cluster representative strategies Westport Crescent City
Comparison 22: Comparing C2, C4, and C6 cluster representative strategies Westport Crescent City
Transect comparisons for C1 and C5 vs. ground truth (all-fine): Westport Crescent City

Scripts to download subsets of the data

Sample_gauges.html shows a notebook that can be adapted to download and plot some synthetic gauge results from this Jupyter notebook: Sample_gauges.ipynb.
For downloading large sets, this script may be easier: download_gauges.py
Sample_hmax.html shows a notebook that can be adapted to download and plot some hmax (maximum flooding depth) results from this Jupyter notebook: Sample_hmax.ipynb.
For downloading large sets, this script may be easier: download_hmax.py