Minutes from 22 April 2009

2009/04/22 CSEP meeting

Attendees: Danijel S., Jeremy Z., Masha L.

1. Status of CSEP V9.4.0 on csep-op
   • No problems
     
2. Tasks for intermediate CSEP V9.4.1 release
   • Trac ticket #147: Assumption in R-test simulation violated
     1. Jeremy submitted a fix in Matlab implementation of R-test to select the same bin in both forecasts
   • Yan's Global long-term and short-term models using CMT
     1. Yan's forecast is using 0.5 degree grid vs. CSEP 1.0 degree grid:
     2. Yan provided conversion routine for short-term forecast file
     3. Will ask Yan to provide routine for long-term forecast file since it has different format
   • Trac ticket #146: Avoid overwriting of modules by testing regions distribution
   • Trac ticket #148: Introduce acceptance tests for web viewer of testing center results
     1. Masha will look at available packages for web site testing and will work with John on their evaluation
        • Python's ClientForm.py to handle HTML forms on the client side
        • HTML parser (BeautifulSoup for example) to parse out returned HTML page from client form submission
        • Python's unittest as testing framework
   • Danijel suggested to fix Trac ticket #145: Change the GMT script for forecast maps to make the maps look like in the RELM report paper
3. John rsync-ed data from current operational server to new hardware with mounted storage
   • Masha will set up crontab with nightly Dispatcher's to mimic operational server to test new hardware
   • Server will be publishing results locally during testing
4. New models
   • Peter Shebalin's EAST three-month alarm-based model
     1. C code compiled on csep-devel
     2. Masha reproduced Peter's reference forecast
     3. Danijel suggested to create new data product: input ANSS catalog not filtered by magnitude - due to increased requests from modelers
     4. Masha created input ANSS catalog that is not filtered by magnitude
     5. Peter created new reference data with that new non-filtered catalog
     6. Masha wrote Python wrapper for the model
     7. EAST is part of the CSEP build
     8. Masha will work on acceptance test for the model
   • Alvaro Gonzalez's one-day alarm-based model
     1. Alvaro met with Masha, discussed CSEP requirements for model submission
     2. Alvaro will have time after May 15 to work on model integration
     3. Might be part of CSEP V9.10.0 if there will be not enough time before code freeze for V9.7.0
5. Jeremy suggested to stop testing STEP model
   • Danijel will contact Matt about Java version of the model
   • Danijel suggested to keep the model unter test for couple of months
6. GEM visit next week
   • Agenda is not known yet