CSEP Minutes 10-03-2018

CSEP Working Group Home Page

• Friday meeting with Ned F. @ USC. Will walk through UCERF3 simulations before running production copies.
• simulations will be used for developing evaluations for stochastic event sets and gathering information
• Talking about forecast horizon for official forecasts
  • New Zealand: 1d, 1w, 3mo, 1yr, etc. based on time since aftershock
  • CEA: interested in short-term and 1yr
• idea: start u3etas in 1985(?) and march forward updating at pre-defined intervals (regular or in-response to earthquake)
• question: how do we perform quantitative comparisons between models?
  • likelihood
  • psuedo-likelihood (moving back to grid)
• immediate: turing tests & u3etas-simulations
• outstanding problems:
  • 1. reproducibility issues within ComCat, should be solved by the USGS
  • 2. comparative testing of models