Predictability study on the aftershock sequence following the 2011 Tohoku-Oki, Japan, earthquake: first results

Abstract

SUMMARY Although no deterministic and reliable earthquake precursor is known to date, we are steadily gaining insight into probabilistic forecasting that draws on space–time characteristics of earthquake clustering. Clustering‐based models aiming to forecast earthquakes within the next 24 hours are under test in the global project ‘Collaboratory for the Study of Earthquake Predictability’ (CSEP). The 2011 March 11 magnitude 9.0 Tohoku‐Oki earthquake in Japan provides a unique opportunity to test the existing 1‐day CSEP models against its unprecedentedly active aftershock sequence. The original CSEP experiment performs tests after the catalogue is finalized to avoid bias due to poor data quality. However, this study differs from this tradition and uses the preliminary catalogue revised and updated by the Japan Meteorological Agency (JMA), which is often incomplete but is immediately available. This study is intended as a first step towards operability‐oriented earthquake forecasting in Japan. Encouragingly, at least one model passed the test in most combinations of the target day and the testing method, although the models could not take account of the megaquake in advance and the catalogue used for forecast generation was incomplete. However, it can also be seen that all models have only limited forecasting power for the period immediately after the quake. Our conclusion does not change when the preliminary JMA catalogue is replaced by the finalized one, implying that the models perform stably over the catalogue replacement and are applicable to operational earthquake forecasting. However, we emphasize the need of further research on model improvement to assure the reliability of forecasts for the days immediately after the main quake. Seismicity is expected to remain high in all parts of Japan over the coming years. Our results present a way to answer the urgent need to promote research on time‐dependent earthquake predictability to prepare for subsequent large earthquakes in the near future in Japan.

A Spatiotemporal Clustering Model for the Third Uniform California Earthquake Rupture Forecast (UCERF3‐ETAS): Toward an Operational Earthquake Forecast

“…In lieu of a detailed review of that model here, Table 1 indicates which types of constraints are embodied in STEP, together with those utilized in the UCERF3 epidemic-type aftershock sequence (ETAS) model presented here. Many other candidate OEF models have also been developed and tested since the introduction of STEP (e.g., Werner et al, 2011;Woessner et al, 2011;Nanjo et al, 2012, and references therein; Segou et al, 2013;Zechar et al, 2013, and references therein;Helmstetter and Werner 2014;Marzocchi et al, 2014;Gerstenberger et al, 2014). What makes UCERF3-ETAS unique to all of these is a more explicit and complete incorporation of geologic fault information, as well as the inclusion of elastic rebound (in which rupture probabilities are thought to drop on a fault after experiencing a large event and to grow back with time as tectonic stresses re-accumulate).…”

Comparative evaluation of physics‐based and statistical forecasts in Northern California

“…It is clear that the rules for forecast model implementation are not the same for physics‐based and ETAS models, but in this study we focus on the comparison between the theoretically advantageous ones from each family, e.g., the forecast time window is set to 10 days for CRS models, but we adopt a daily updated ETAS model, since Nanjo et al . [] has demonstrated the link between poor performance and extended update intervals for statistical models.…”

“…We also compare seven ETAS realizations; those are based on two basic principles of ETAS modeling that (1) the update time should be reasonably short (24 h) and (2) all available earthquakes should be used for cascading earthquake sequences, which is found to improve statistical forecasting methods [Helmstetter, 2003]. It is clear that the rules for forecast model implementation are not the same for physics-based and ETAS models, but in this study we focus on the comparison between the theoretically advantageous ones from each family, e.g., the forecast time window is set to 10 days for CRS models, but we adopt a daily updated ETAS model, since Nanjo et al [2012] has demonstrated the link between poor performance and extended update intervals for statistical models.…”

Prospective Earthquake Forecasts at the Himalayan Front after the 25 April 2015M 7.8 Gorkha Mainshock

“…Neither method flawlessly accounts for the spectrum of post mainshock earthquake behavior (e.g., Nanjo et al, 2012;Cocco et al, 2010;Segou et al, 2014). Empirical/statistical models depend heavily on the density of the local seismic network and identification of precursory activity and are thus vulnerable in frontier regions.…”