SCEC Award Number 13183 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title VISES Travel Proposal:Comparison of Short-Term Earthquake Predictability in California and Japan
Investigator(s)
Name Organization
Maximilian Werner Princeton University
Other Participants Prof. Naoshi Hirata, ERI, Univ. Tokyo
SCEC Priorities 2b, 2e, 4e SCEC Groups CSEP, EFP, Seismology
Report Due Date 03/15/2014 Date Report Submitted N/A
Project Abstract
 The objectives of this VISES project were to calibrate earthquake forecasting models developed for California to Japan. The goal is a deeper understanding of the predictability of earthquakes in different tectonic environments. PI Max Werner spent two weeks at the Earthquake Research Institute in Tokyo in 2013 to collaborate with Professor Hirata and members of his research group, which form the core of the CSEP testing center in Japan. Initial plans to study the short-term predictability of earthquakes were amended in light of the focus on intermediate-term forecasts at the Japanese testing center. Dr. Tsuruoka provided earthquake catalog data from the Japanese Meteorological Agency (JMA) to calibrate the adaptive space-time smoothing models developed by Helmstetter and Werner (2012). We calibrated the models to the three Japanese testing regions, which are the Kanto region, the onshore mainland area, and a wider region including the offshore subduction zones. To test the sensitivity of the estimated parameters and the probability gains, we used three different calibration time windows, two spatial smoothing kernels and two different density estimation methods. Our results suggest interesting differences in both estimated parameters values and retrospective probability gains. Parameter values take a broader range and can vary substantially across testing regions; gains are almost universally smaller than in California's transform boundary environment. Results of this study were presented at the SCEC annual meeting in 2014.
Intellectual Merit The Japanese CSEP testing region is substantially more seismically active than California and it covers a different tectonic setting. Our results are beginning to address the questions to what extent earthquake predictability depends on tectonic setting and whether earthquake forecasting models can be easily exported to other regions, e.g. for purposes of operational earthquake forecasting.
Broader Impacts CSEP is a global initiative that relies on communication, open data exchange and collaboration. This proposal facilitated the visit of PI Max Werner to the Japanese CSEP node and thereby strengthened the level of interaction between the SCEC and ERI CSEP testing centers. Werner disseminated CSEP SCEC results to the ERI and wider seismological community in Tokyo. Frequent meetings between Werner and members of Hirata's group led to new insights into CSEP operations and experiments. Werner helped train postdoc Yokoi in CSEP concepts and testing methods.
Exemplary Figure Figure 3: Left: Pseudo-prospective forecast map for the 3-month period starting 1 February 2011 and including the M9 Tohoku earthquake using the power-law Conan model with parameters optimized during W_1. Right: Comparison of observed and predicted numbers of earthquakes over multiple 3-month periods since the start of the CSEP Japan 3-month forecast group.