Results of Call for earthquake forecast models, test regions

1
Results of Call for earthquake forecast models,
test regions forecast evaluation methods
International symposium Toward constructing
earthquake forecast systems for Japan 27 May
2009 at ERI, Univ. Tokyo

• Research group of earthquake forecast systems
  based on seismicity of Japan in collaboration
  with D. Schorlemmer, F. Euchner, and CSEP working
  group

2
The earthquake forecast testing experiment

• Current stage Testing Center was installed and
  three models are under test.
• Next stage Use the existing Testing Center to
  start a formal testing this academic year

Triple-S
RI
ALM
3
Rules of the game

• Testing regions
• Default regions (d30km) with grid spacing 0.1
  deg.
• Invite test regions
• Testing classes (Invite forecast models)
• 1-day forecast
• 1-year forecast
• 5-year forecast
• Future earthquakes that will be forecast
• The revised JMA bulletin. No declustering.
• Number of earthquakes in a predefined time window
  for each magnitude bin in the range 5.0M9.0
  (0.1 magnitude unit steps) at each predefined
  grid node within a predefined testing region.
• Forecast evaluation methods
• CSEP official suite
• Invite forecast evaluation methods
• We noticed this experiment to seismologists
  through our webpage and Newsletter of Seism. Soc.
  Japan in 2009

4
Give notice of proposals

• Earthquake forecast model proposal
• Test region proposal and applicable forecast
  models
• Forecast evaluation method proposal

5
Forecast models

• Seismicity-based forecast models with different
  earthquake-generation hypotheses.
• ALM D. Schorlemmer (1 yr)
• Triple-S J. Zechar (1 yr)
• RI K.Z. Nanjo (1 yr, 1 day if there is any
  proposed one-day model)
• Asperity-based S. Matsumura (1 yr)
• Double-branching W. Marzocchi (1 yr need to be
  checked)
• ETAS Ogata (1 yr need to be checked)
• ETAS J. Zhuang (1 yr need to be checked)
• MARFS, MARFSTA, JMM C. Smyth (1 yr)

6
Proposed test regions
Default all Japan with d30km (proposed by CSEP)
ALM , Triple-S, RI, Double-branching, ETAS
(Ogata), ETAS (Zhuang), MARFS, MARFSTA
JMM. Same with d100km (proposed by EFSSJ)
Models are invited
Off Ibaraki with d100km (proposed by Matsumura)
Asperity-based
Kanto with d100km (proposed by Nanjo) Models
are invited
7
Evaluation methods

• CSEP official suite that is currently used J.
  Zechar
• Likelihood-based N, L, R-tests
• Alarm-based tests
• Conventional likelihood-based methods M. Imoto