Towards a harmonized operational earthquake forecasting model for Europe

Abstract

We develop a harmonized earthquake forecasting model for Europe based on the epidemic-type aftershock sequence (ETAS) model to describe the spatiotemporal evolution of seismicity. We propose a method modification that integrates information from the 2020 European Seismic Hazard Model (ESHM20) about the spatial variation in background seismicity during ETAS parameter inversion based on the expectation–maximization (EM) algorithm. Other modifications to the basic ETAS model are explored, namely fixing the productivity term to a higher value to balance the more productive triggering by high-magnitude events with their much rarer occurrence and replacing the b-value estimate with one relying on the b-positive method to observe the possible effect of short-term incompleteness on model parameters. Retrospective and pseudo-prospective tests demonstrate that ETAS-based models outperform the time-independent benchmark model as well as an ETAS model calibrated on global data. The background-informed ETAS variants using the b-positive estimate achieve the best scores overall, passing the consistency tests and having a good score in the pseudo-prospective experiment. Our test results also highlight promising areas for future exploration, such as avoiding the simplification of using a single b value for the entire region, reevaluating the completeness of the used seismic catalogs, and applying more sophisticated aftershock spatial kernels.