CSEP Minutes 06-27-2018

CSEP Working Group Home Page

Participants: M. Werner, D. Rhoades, and W. Savran

Minutes

• extract daily rates for each forecast and plot to verify intregity (first step)
• verify where W, T and R tests are not counted for entire time period
• decipher the timestamp on the evaluation file
• how does the system put together the cumulative tests
• are there cumulative analogs for each evaluation?
• curated csep1.0 data might be needed to compare against csep2.0 forecasts
• verification exercises:
  • overlapping window: ETAS extract first month of forecasts and observed catalogs and evaluations
    • RELM tests (N, L, CL, M, S)
  • start at the beginning of time to verify cumulative tests
  • write scripts for this
• scripts to extract evaluation results
• journal for publishing geoscience data (https://www.earth-system-science-data.net/)
• check existence of the catalog data
  • we would need to flag evaluations where catalog could not be downloaded
• competition for fault-based forecasts?
  • some hazard models have the notion of seismic regions
  • new zealand has the issue of faults
  • ucerf3 most advanced
  • WGSEP88 report model (D. Jackson paper in the SRL issue; good example for assessing fault based source model)

Action Items

Bill:

• provide data set to verify forecasts by extracting the daily-rates for each forecast.
• provide secondary data set to verify evaluations, which consists of forecasts and observations needed and the evaluations to compare against.
• double check run-time estimates for CSEP1 reprocessing estimates.
• generate list of days where observations are missing