Difference between revisions of "Basin Query Notes"
(Created page with "I have created all of the plots discussed (and a few more). I did this by creating a new OpenSHA class, ZFilePlot. That class is posted here: https://github.com/opensha/opens...") |
|||
| Line 1: | Line 1: | ||
| − | + | Kevin created all of the plots discussed (and a few more). I did this by creating a new OpenSHA class, ZFilePlot. That class is posted here: | |
https://github.com/opensha/opensha-cybershake/blob/master/src/scratch/kevin/cybershake/ZFilePlot.java | https://github.com/opensha/opensha-cybershake/blob/master/src/scratch/kevin/cybershake/ZFilePlot.java | ||
Revision as of 22:16, 27 November 2017
Kevin created all of the plots discussed (and a few more). I did this by creating a new OpenSHA class, ZFilePlot. That class is posted here:
A build of this tool, along with the bash script and all data I used to run it is posted here:
http://opensha.usc.edu/ftp/kmilner/cvm_data/2017_z_file_compare/
The usage of the tool (requiring java 8 and after downloading the jar file) is:
java -cp opensha-cybershake-all.jar scratch.kevin.cybershake.ZFilePlot <min-lat> <min-lon> <nx> <ny> <grid-spacing> <max-z> <data-file> <label> [<compare-file> <compare-label>]
The "max-z" argument scales the color pallet to this value (I use 10 for Z2.5 and 2 for Z1.0). All plots are in km (the files themselves are in meters, but I scale them before plotting). The last 2 arguments are optional, for creating ratio/scatter plots. For example, to compare basin_query Z2.5 to OpenSHA Z2.5:
java -cp opensha-cybershake-all.jar scratch.kevin.cybershake.ZFilePlot 33.35 -123 1551 1201 0.005 10 basin_query_mpi_cca_z2.5_n20_second.result BQ\ CCA\ Z2.5\ Second opensha_cca_second_z2.5.bin OpenSHA\ Z2.5\ CCA\ Second
The forward slashes in the label arguments are for spaces within a single argument. While I hope this tool will be useful to Mei in the short-term, I don't believe that it should be bundled with UCVM (it relies on an opensha web service that may change/move at any time), and I still believe that a python/matplotlib plotting script should be created within UCVM.
Plots are posted here (I'll let Phil and Mei decide which plots are desired for the wiki, as uploading all of them manually is rather onerous):
http://opensha.usc.edu/ftp/kmilner/cvm_data/2017_z_file_compare/plots/
For example:
basin_query Z2.5 second crossing: 
OpenSHA Z2.5 second crossing: 
Log-ratio of basin_query to OpenSHA: 
Scatter plot of basin_query vs OpenSHA: 
Those plots do show a difference in the registration of the model in Mei's files. This is probably related to the issue Christine described.
A zip file with the code, all data files, and all plots is posted here as that may be more convenient than manually downloading everything:
http://opensha.usc.edu/ftp/kmilner/cvm_data/2017_z_file_compare.zip
