Minutes from 8 October 2008

2008/10/08 CSEP meeting

Attendees: Yan K., Phil M., Jeremy Z., Danijel S., Matt G., John Y., Masha L.

1. Status of CSEP V8.10 release:
   • Running successfully on csep-op since installation
   • RAM is upgraded, reboot didn't prompt for the disk check since it was done during last reboot
   • Phil will write a summary for the release announcement
2. ACES paper is complete, Jeremy will submit once final draft is reviewed by Tom
3. Kagan-Jackson Smooth Seismisity model for California:
   • One-day and 5-year models
   • 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. Change parameter value as it had to be done for WP regions to get matching long-term forecast
     3. Short-term forecast is not matching - have to provide time of the run through parameter file instead of using date and time of the run
   • Yan asked if we can share the modified code with Italy
   • Yan pointed out that model region is 500 km larger vs. RELM collection area
     
     1. Danijel and Yan agreed to use RELM collection area for input catalog passed to the model
4. Upgrade of New Zealand testing center to CSEP V8.10:
   • To be addressed
     
     1. Linux distribution
     2. CSEP V8.10 and all external packages CSEP V8.10 is dependent on

     3. QuakePy V0.1.1

   • Will get development machine to:
     
     1. Build release candidates
     2. Install new models

     3. Masha introduced GeoNetNZDataSource.py module to automatically download catalog in XML format, and to pre-process into ZMAP format using QuakePy

   • QuakePy remaining issues:

     1. Exception when processing catalog that starts with 01/01/1863
   • Danijel will work on GMT script for forecasts maps
   • Grid shifting - Danijel suggested to use global grid notation
   • Existing CSEP method of results publishing to be used: rsync
   • Web-viewer for the results, still to be resolved to use:
     
     1. Stand-alone web page or
     2. Introduce as new testing region within existing CSEP results page
5. Catalog declustering:
   • Should schedule Fortran code review mid-November to consider:
     
     1. Use within CSEP
     2. Re-write in Python
   • Jeremy suggested to consider other declustering methods
   • Matt is using Matlab version of declustering for New Zealand
   • Wait with intermediate release that addresses use of true declustered catalog within RELM Mainshock groups
     
     1. May be until CSEP V9.1.0