Exciting news! We're transitioning to the Statewide California Earthquake Center. Our new website is under construction, but we'll continue using this website for SCEC business in the meantime. We're also archiving the Southern Center site to preserve its rich history. A new and improved platform is coming soon!

Poster #254, Earthquake Forecasting and Predictability (EFP)

Towards next-generation earthquake forecasting by embracing short-term aftershock incompleteness

Leila Mizrahi, Shyam Nandan, & Stefan Wiemer
Poster Image: 

Poster Presentation

2021 SCEC Annual Meeting, Poster #254, SCEC Contribution #11481 VIEW PDF
Epidemic-Type Aftershock Sequence (ETAS) models are the most successful earthquake forecasting models currently available, both for short- and long-term hazard assessment. They account for the spatio-temporal clustering of earthquakes intrinsically using basic empirical triggering laws. It is thus reasonable to build on top of these simple yet effective models when developing next-generation earthquake forecasting models. Additional knowledge about earthquake physics could be incorporated into existing models through the effective use of Coulomb stress changes, spatial variations of the b-value, or anisotropic triggering based on fault geometry. On the other hand, current models can be enhan...ced to leverage the growing amount of data made available by continually improving seismic networks.

We examine several promising approaches to developing next-generation earthquake forecasting models, with particular emphasis on how to embrace data incompleteness in the form of short-term aftershock incompleteness (STAI) when using ETAS. We designed a self-consistent algorithm to estimate high-frequency detection incompleteness and ETAS parameters jointly. For this, we generalized the concept of completeness magnitude mc and consider a rate- and magnitude-dependent detection probability. Using pseudo-prospective forecasting experiments in California, we tested the usefulness of the different components of this new model for forecasting, and plan to test them within upcoming CSEP experiments. In this way, we aim to shed light on which aspect of the model is the most promising to pursue on the way to next-generation earthquake forecasting.

Preliminary results indicate that modelling aftershocks of small magnitude events leads to significantly improved forecasts compared to the current state-of-the-art base model, but only if short-term aftershock incompleteness is estimated and accounted for. This improvement decreases when the difference in magnitude between forecasted and newly included earthquakes increases, which can potentially be explained by previous studies which found that earthquakes tend to trigger similarly sized aftershocks.