Minutes from 20 August 2008

2008/08/20 CSEP meeting

Attendees: Yan, Jeremy, John, Masha

1. Western-Pacific testing regions:
   • Double Branching model for NW Pacific is fully automated within CSEP:
     
     1. Created DBMOneYearModel Python wrapper class
     2. Added acceptance test that is part of nightly CSEP build:
        • Anna Maria updated Fortran code so CSEP can specify if random seed value is read from the file or drawn by the system
     3. Created new one-year forecast group with later entry date (August 20, 2008) into the testing center on csep-cert:

        • Includes Triple_S, KaganJacksonSmoothSeismisity, and DoubleBranchingModel

        • Invokes N, L and R tests for the models
     4. Updated configuration file used by nightly crontab to invoke and to evaluate the models
   • Fixed Trac ticket # #89: Automate TODAY to UTC adjustment in Yan's forecasts:
     
     1. CSEP calculates current offset in hours and passes the value to the model through input parameter file
2. Software development:
   • Installed V8.7.1 on operational machine Thursday, August 7, 2008
     
     1. Re-processed latest test date for RELM forecasts groups to display cumulative R-tests results with proper labels
     2. Includes fixes for Trac tickets:
        • #98 Re-processing of the test date doesn't stage existing modifications catalogs
        • #93 Change of CMT web-site directory structure causes extra files being downloaded
        • #97 Reduce disk usage by removing originally downloaded and pre-processed catalogs files
        • #96 Cumulative R-test result plot should switch curves labels
        • #95 Should not scan forecast directory for staged files at construction time
        • #94 Use CSEPStorage functionality to stage archived XML format forecast file if it exists
   • Version 8.10.0 goals:
     
     1. Double Branching Model for NW and SW Pacific testing regions:
        • NW Pacific is done
        • Waiting for an example of input parameter file and all required input files for SW Pacific testing region from Anna Maria
     2. Jeremy put Java code on motion and sent Masha e-mail with instructions on how to build and run the code
        • Alarm-based versions of TripleS forecasts
        • Alarm-based tests
     3. Possibly Kagan-Jackson Smooth Seismicity Model for California

        • Yan pointed to newly available documents on http://scec.ess.ucla.edu/~ykagan/calnev_index.html

        • Yan will send e-mail when forecast source code is posted on the web-site
   • Max Werner has asked for an example of input catalog - don't know when he plans to install the model within CSEP
3. Masha submitted a poster abstract for the SCEC 2008 Annual Meeting