Science of OEF Meeting
Meeting Goals:
Building on our Potential Uses of OEF meeting and associated report, the goal of this meeting is to discuss and identify best available science (in the context of likely uses), and to construct a prioritized implementation plan for a California OEF.
Background and Meeting Strategy:
Having surveyed the range of possible OEF uses at a previous Powell Center meeting, we now move on to identifying the best available science, both for nearterm deployments and for more advanced capabilities anticipated in the future. Because all models are only an approximation of the system, best available science must be ascertained in the context of specific uses. Also, and as emphasized at our previous meeting, it is important to identify and work directly with a set of early adopters, which for our planning purposes can be considered as follows:

CEPEC and NEPEC

CalEMA

PG&E

California Earthquake Authority (perhaps through the commercial loss modeling companies they rely upon)
Table 1, on the following page, lists various potential OEF products, ranging from a magnitudeprobability distribution, to a full earthquakerupture forecast, to various risk estimates.
Because OEF aims to “to help communities prepare for potentially destructive earthquakes” (Jordan et al., 2014), usefulness will be dictated by reliability and skill at larger magnitudes (e.g., M≥6.5 in California). Table 2 lists various informational constraints that could be utilized in forecasting such events. As indicated therein, an explicit goal of UCERF3ETAS has been to incorporate finite faults, elastic rebound, and characteristic magnitude frequency distributions. Interestingly, we find that you cannot combine ETAS with finite faults without also including elastic rebound, and adding the latter in turn implies a need for characteristic magnitudefrequency distributions on faults. Considerable time will be spent discussing UCERF3ETAS at this meeting, especially given the complexity of the model. However, we will also review and discuss both simpler and more advanced approaches (e.g., physicsbased simulators).
We currently plan to have two follow up Powell Center meetings, one to address operationalization challenges and one on verification and validation.
Table 1. Potential products that could be produced either directly by an operational earthquake forecast system, or by further processing in downstream risk calculations. Each of these will constitute an estimate in that uncertainties will need to be quantified (taken from the Potential Uses of OEF report).
Product/Metric 
Description 
Magnitude Probability Distribution 
The likelihood of having earthquakes of different magnitudes in a given area and for a specified timespan. 
Spatial Distribution of Triggering 
Specifying where triggered events are most likely to occur. 
Fault Triggering Probabilities 
A prioritized list of faults that could nucleate or participate in large triggered events, accounting for any elasticrebound effects. 
Full Earthquake Rupture Forecast (ERF) 
Specifying the likelihood of all possible events (at some discretization level) for a given timespan. 
Stochastic Event Sets 
Synthetic catalogs of triggered events as implied by an ERF. 
Hazard Estimates 
The probability of exceeding hazardrelated intensity measures (e.g., PGA, PGV, PGD, SA, MMI) as a function of time and space. 
Sequence Duration 
Time needed for some measure (e.g., earthquake rate) to drop back to some level 
Scenario Earthquakes 
Representative examples of earthquakes that could be triggered 
Ground Deformation Probabilities 
Forecast of future fault offsets (e.g., due to creep) and/or other types of ground deformation 
Landslide Probabilities 
The likelihood of triggering landslides 
Liquefaction Probabilities 
The likelihood of triggering liquefaction 
 HazardRisk Separation Interface  

Population and/or Infrastructure Exposed 
The number of people, houses, commercial properties, schools, hospitals, etc. with a certain likelihood of experiencing certain shaking or other hazard thresholds 
Deaths/Injuries/Hospitalizations 
Risk estimates with respect to human physical health and survival. 
Damage Level and Collapse Probability 
Loss and collapse estimates with respect to built infrastructure. 
Downtime 
Likelihood and length of disruption to business, power, water, waste disposal, telecom, and communication systems. 
Inspection Priority or Concern Level 
A customized and prioritized list of assets that may require attention (e.g., as provided by the ShakeCast system). 
Table 2. Some of informational constraints (observed or model inferred) that one might use to forecast the probability of triggered earthquakes (taken from the UCERF3ETAS documentation).
Observation or Model Inference 
STEP 
UCERF3ETAS 
Global or regional triggering statistics (generic aftershock parameters) 
√ 
√ 
Sequencespecific deviations from generic aftershock parameters 
√ 
√ 
Spatial and temporal variation in aftershock parameters (within a sequence) 
√ 

Spatial variation in longterm event rates (background rates) 

√ 
Location of recent events (i.e., areas lighting up with microseismicity) 
√ 
√ 
Proximity to active faults, especially with respect to triggering largermagnitude events 

√ 
The longterm magnitudefrequency distribution inferred for faults 

√ 
Elasticrebound implied stress on faults (e.g., time since last event relative to recurrence interval) 

√ 
Dynamic or static stress changes imposed by any previous events 
Implicit? 
Implicit? 
Transient deformation 


Meeting Participants
Name 
Affiliation 

Blanpied, Michael 
USGS, Reston; NEPEC 
mblanpied@usgs.gov> 
Dieterich, James 
UC, Riverside 
james.dieterich@ucr.edu> 
Fialko, Yuri 
UC, San Diego 
yfialko@ucsd.edu> 
Field, Ned 
USGS, Golden 
field@usgs.gov> 
Gerstenberger, Matt 
GNS New Zealand 
m.gerstenberger@gns.cri.nz> 
Gilchrist, Jacquelyn 
UC, Riverside 
jacquelyn.gilchrist@email.ucr.edu> 
Gomberg, Joan 
USGS, Seattle 
gomberg@usgs.gov> 
Hardebeck, Jeanne 
USGS, Menlo 
jhardebeck@usgs.gov> 
Jordan, Thomas 
USC 
tjordan@usc.edu> 
Michael, Andrew 
USGS, Menlo; NEPEC 
michael@usgs.gov> 
Milner, Kevin 
USC 
kmilner@usc.edu> 
Page, Morgan 
USGS, Pasadena 
mpage@usgs.gov> 
Powers, Peter 
USGS, Golden 
pmpowers@usgs.gov> 
Rhoades, David 
GNS New Zealand 
D.Rhoades@gns.cri.nz> 
RichardsDinger, Keith 
UC, Riverside 
keithrd@ucr.edu> 
Shaw, Bruce 
LDEO 
shaw@ldeo.columbia.edu> 
Trugman, Daniel 
UCSD grad student 
dtrugman@ucsd.edu> 
van der Elst, Nicholas 
USGS, Pasadena 
nicholas.vanderelst@gmail.com> 
Werner, Maximilian 
U. Bristol 
max.werner@bristol.ac.uk> 