Home » News » 2008 » Minutes from 19 November 2008

Minutes from 19 November 2008

2008/11/19 CSEP meeting

Attendees: Max W., Fabian E., Danijel S., Jeremy Z., John Y., Masha L.

  1. csep-op:
    • There were failures for two days due to Matlab license problems
    • These days were not reprocessed yet
  2. Double Branching Model from Anna Maria for Western Pacific testing regions:
    • Masha successfully generated SW Pacific forecast with 10000 simulations on csep-devel server:
      1. Forecast matched Anna Maria's forecast
    • Updated source code and reference data used by acceptance tests for the model in repository
    • Nightly build ran model's acceptance tests successfully
    • Masha introduced one-year-models-V9.1 forecasts groups for SW and NW Pacific regions on csep-cert server:
      1. Includes Yan's, Jeremy's and Anna Maria's one-year forecasts
      2. Invokes N, L, R, ROC, Molchan and ASS tests
      3. Tonight's crontab jobs should generate forecasts with start date of 10/20/2008
  3. Upgrade of New Zealand testing center to CSEP V8.10:
    • Masha is working with Matt over Skype
    • 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

        • John is helping to set up passwordless connection for csep-nz user to cseptesting.org to publish results under /var/www/html/csep/nz/results/data/nz/gns directory

  4. Wish list from Jeremy:
    • Test date of July 1, 2008, was reprocessed on csep-cert for RELM Mainshock groups:
      1. N, L, and R tests
      2. Declustered catalog is used as observation catalog
    • Masha is generating L-test data to test it's stability:
      1. RELM Mainshock and Mainshock/Aftershock groups
      2. Staging observation catalog per model as provided by Jeremy
  5. ETAS one-day model installation within CSEP:
    • Max sent an email to Matt and Zhuang asking if they would agree to reduce magnitude threshold to 2.0 vs. 3.0 currently used by the CSEP:
      1. So far no response
    • Danijel suggested to remove magnitude filtering from catalog since there are different requests for magnitude threshold from different modelers
    • Max currently generates forecasts real-time, and evaluates them the following day at mid-night:
      1. CSEP is using 31-day delay which was not investigated if it's sufficient
      2. ANSS suggested 2-weeks delay
    • Max would like to submit two versions of ETAS for testing:
      1. Masha asked for the model to support files locations for results forecasts and input catalog
      2. Model is not using random numbers
      3. Max will send an example of script that invokes series of different codes (Fortran 77, Matlab) to generate the forecasts
      4. Max should have code ready for the CSEP early next week
  6. S("space") evaluation test to be added in V9.1:
    • Jeremy is working on Matlab code, will send Masha once it's complete
    • Masha will introduce Python class for the test
    • Should try to invoke the test for all existing forecasts groups on csep-cert server
  7. Fixes for ANSS catalogs to be implemented:
  8. Global testing:
    • Data source: CMT
    • No collection or test area
    • Danijel will work on defining ForecastML template
    • Jeremy and Danijel will submit two flavors of Triple-S model for testing
Home » News » 2008 » Minutes from 19 November 2008