Minutes from 16 April 2008
2008/04/16 CSEP meeting
Attendees: Yan, Jeremy, John, Amit, Masha
- Introducing Western Pacific testing region into the CSEP:
- Yan confirmed that forecasts converge
- Yan recommended to check now and then if forecasts converge, let Yan know if there are any issues
- Masha asked about time duration for long-term and short-term forecasts:
- Yan suggested to use 1 year for long-term, and 1 day for short-term forecasts for now
- Will discuss details in person during CSEP Global Collaboration Meeting on Monday, April 21, 2008
- Amit is working on Python class to automate Yan's forecasts within CSEP
- Need to have Python class for pre-processing of input catalog for the model:
- Need to convert pre-processed (1-line per event) CMT catalog to Yan's expected binary format
- Jeremy suggested to have it as one of the forecast invocation steps
-
Since input catalog is shared by all models within forecast group, might need to have generic filtering to be applied within that class.
Currently, for all installed models within CSEP input catalog preparation consists of:- Cut to magnitude
- Cut to collection area
- Cut to time period
-
Masha will work on the catalog preparation class once testing of new BatchProcessing module is complete
- Need to convert pre-processed (1-line per event) CMT catalog to Yan's expected binary format
- Software development:
-
BatchProcessing module is complete
- Masha is testing locking of forecasts directories when multiple processes use the same forecast group
- Need to prioritize tasks for the next Version 8.7.0 release:
- Introduce Yan's models for Western-Pacific testing region
- Warner's model from Italy for Western-Pacific testing region
- Uncertainties for cumulative results
- Combine intermediate and true cumulative results in one plot
- Adopt QuakeML format for catalog data
-
- CSEP Global Collaboration and Testing Meeting on April 21, 2008:
- Jeremy suggested to show results web-page during software presentation if nobody else is going to present it
