Minutes from 15 October 2008

2008/10/15 CSEP meeting

Attendees: Jeremy Z., John Y., Masha L.

1. Status of CSEP V8.10 release:
   • Running successfully on csep-op since installation:
     
     1. Last night ran alarm-based tests for one-year forecasts groups of Western Pacific that is scheduled for 14 and 28 day of each month
2. Kagan-Jackson Smooth Seismisity model for California:
   • Masha downloaded source code, compiled with ifort without any changes
   • To invoke forecast, Masha had to make modifications:
     
     1. Change hard-coded path to input catalog
     2. Provide start date and time for forecast through parameter file instead of obtaining it at run-time
     3. Long-term and short-term forecasts are matching with less than 0.5% differences
   • Yan uses ANSS catalog format as downloaded through the web-site
   • CSEP is using ZMAP format of ANSS catalog:
     
     1. Masha will send email to Yan asking if Yan will make code changes to read input catalog in ZMAP format
   • Jeremy confirmed that model is not using focal mechanisms from catalog, ZMAP format should provid all parameters the model is using
3. Upgrade of New Zealand testing center to CSEP V8.10:
   • Accomplished:
     
     1. Linux distribution didn't need to be changed
     2. Installed all external packages CSEP V8.10 is dependent on

     3. Installed QuakePy V0.1.1

     4. Masha introduced trunk/src/SCECModels/NewZealand/generic/GeoNetNZDataSource.py module to automatically download catalog in XML format, and to pre-process into ZMAP format using QuakePy:

        • Start date of 01/01/1863 for GeoNet catalog causes malformed XML format file resulting in QuakePy exception:

          1. The earliest start date is 01/01/2004 that does not cause the problem
     5. Masha updated configuration files for the center
     6. Matt ran the build for V8.10 code distribution:
        • Failed collecting versions of existing packages
          
          1. Please see Trac ticket #112 - fix is implemented not to hard-code packages within CSEPStatus.software() method, but to allow each installed forecast model to report the commands to collect versions of packages they are dependent on
   • Matt will be available some time next week to install New Zealand specific codes
   • Remaining tasks:
     
     1. Danijel will work on GMT script for forecasts maps
     2. Grid shifting - Danijel suggested to use global grid notation
     3. Web-viewer for the results, still to be resolved to use:
        • Stand-alone web page or
        • Introduce as new testing region within existing CSEP results page

4. Masha reproduced null probability error for Double Branching Model of South Western Pacific testing region on new csep-devel integration server:

   • Problem does not show up on motion, but on csep-cert server
   • Sent e-mail to Anna Maria asking to use csep-devel to trace the problem