SCEC Award Number 17175 View PDF
Proposal Category Workshop Proposal
Proposal Title CSEP workshop: Informing earthquake debates with CSEP results
Name Organization
Maximilian Werner University of Bristol (United Kingdom) David Rhoades GNS Science (New Zealand) Andrew Michael United States Geological Survey Thomas Jordan University of Southern California Warner Marzocchi Istituto Nazionale di Geofisica e Vulcanologia (Italy)
Other Participants
SCEC Priorities 5a, 5b, 1b SCEC Groups EFP, CSEP, WGCEP
Report Due Date 10/09/2017 Date Report Submitted 11/14/2017
Project Abstract
The Collaboratory for the Study of Earthquake Predictability (CSEP) aims to develop a global cyberinfra-structure for the prospective and blind evaluation of earthquake forecasting models and prediction algo-rithms. CSEP thereby contributes to the objective assessment of the predictive power of scientific hy-potheses about earthquake occurrences. This workshop had the following objectives: 1) to retrieve and interpret new CSEP results from around the globe, 2) to develop a publication plan, 3) to plan next steps for retrieving short-term model results, and 4) to wrap phase 1 of CSEP and begin planning a phase 2 that is aligned with current community needs. A volume with extended abstracts is available on request. Speakers presented CSEP results from California, New Zealand, Italy, Japan and the global experiment. Participants agreed to publish new CSEP results in a focus section of the Seismological Research Let-ters (submission deadline 1 February 2018 and publication date Jul/Aug 2018). Wrapping up phase 1 of CSEP will require reprocessing in California and New Zealand to obtain complete evaluations of short-term models because of earthquake catalog issues (in New Zealand) and incompletely processed results. Recommendations were collected for CSEP2.0, ranging from new experiments and tests (e.g. to test UCERF3-ETAS) to recommended model developments to accessible outreach tools.
Intellectual Merit The results contribute to SCEC’s goal of understanding the predictability of earthquakes. Workshop results suggest that strain-rate based forecasts are competitive alternatives to smoothed seismicity forecasts; Coulomb-based fore-casts can compete with purely statistical models; and new ways for assessing and visualizing predictive skills have been developed.
Broader Impacts The predictability of earthquakes is of broad interest. Government agencies use seismic hazard models for building planning and other purposes, but the underlying hypotheses in source models remain debated. Our results contrib-ute to this debate. Early career researchers (including two women) were invited to present their work. SCEC-sponsored CSEP workshops remain the global focal point for CSEP collaborations and progress.
Exemplary Figure Figure 1: Spatial visualization of the likelihood of earthquakes M4.95+ between January 2011 and August 2017 given the probabilistic forecast “Neokinema” by Bird & Liu (2007). Size of squares (earthquakes) indicates magnitudes; in-fill colour of squares denotes difference in log-likelihood between Neokinema and a spatially uniform benchmark. The Brawley earth-quakes of 2012 are well forecast by Neokinema; the Hawthorne earthquakes in 2016 are not, and two more earthquakes are surprising (circles).