Difference between revisions of "CSEP Minutes 10-24-2018"
From SCECpedia
Jump to navigationJump to search (Created page with "CSEP Working Group Home Page<br><br> ''Participants'': William Savran, Andy Michael, David Rhoades, Jacqui Gilchrist, Warner Marzocchi, Max Werner, Ann...") |
|||
| (One intermediate revision by the same user not shown) | |||
| Line 6: | Line 6: | ||
= Minutes = | = Minutes = | ||
| + | The topics of the call were about the current California aftershock forecasting experiment at the SCEC Testing Center and developing evaluations for simulations that produce stochastic event sets. | ||
== Simulations == | == Simulations == | ||
| − | + | * Verify with Ned: Does no faults ETAS include characteristic earthquake behavior? If a large earthquake does not rupture on a known fault does that produce an isotropic aftershock distribution. | |
* Characteristic EQ Behavior: the larger earthquakes in a segment are overrepresented wrt to their extrapolation in a GR plots. | * Characteristic EQ Behavior: the larger earthquakes in a segment are overrepresented wrt to their extrapolation in a GR plots. | ||
* The aftershocks still follow GR (in ETAS), so the larger proportion of large earthquakes causes more aftershocks and can lead to runaway sequences. | * The aftershocks still follow GR (in ETAS), so the larger proportion of large earthquakes causes more aftershocks and can lead to runaway sequences. | ||
Latest revision as of 20:27, 24 October 2018
Participants: William Savran, Andy Michael, David Rhoades, Jacqui Gilchrist, Warner Marzocchi, Max Werner, Anne Strader, Dave Jackson, Phil Maechling
Contents
Minutes
The topics of the call were about the current California aftershock forecasting experiment at the SCEC Testing Center and developing evaluations for simulations that produce stochastic event sets.
Simulations
- Verify with Ned: Does no faults ETAS include characteristic earthquake behavior? If a large earthquake does not rupture on a known fault does that produce an isotropic aftershock distribution.
- Characteristic EQ Behavior: the larger earthquakes in a segment are overrepresented wrt to their extrapolation in a GR plots.
- The aftershocks still follow GR (in ETAS), so the larger proportion of large earthquakes causes more aftershocks and can lead to runaway sequences.
- David Rhoades simulation-based model can produce stochastic event sets. With little modifications.
- How long will it take until we know whether or not faults make the difference in the simulations?
- Can deal with fault-finiteness by pulling from the ComCat or looking at earthquake aftershocks
Evaluations
- Warner and David supplied a document outlining the new possible tests for CSEP2 simulation- based tests.
- Tests typically follow CSEP1 consistency testing ideas with some emphasis on formulating likelihood based tests.
- Need to put some more effort into understanding likelihood (comparative) based-testing.
- Paper on proper scoring: https://www.stat.washington.edu/raftery/Research/PDF/Gneiting2007jasa.pdf
