Using ETAS+ to forecast aftershock sequences

Nicholas J. van der Elst

Published September 8, 2024, SCEC Contribution #13953, 2024 SCEC Annual Meeting Poster #211

Earthquake catalogs provide the core data of statistical seismology but are typically incomplete records of the earthquakes that occur. Incomplete detection results from sparse network coverage and saturation of the network during periods of high activity. Using the ‘positive’ statistics approach, statistics are estimated only within complete intervals defined between pairs of earthquakes where the second is larger than the first. The first earthquake in the pair sets a reference magnitude above which subsequent earthquakes are likely to be detected, and a small positive magnitude adjustment ensures that the next earthquake is detected with arbitrarily high probability. Here I extend the approach to the epidemic-type aftershock sequence (ETAS) model. For every earthquake, I define a complete magnitude-time interval terminated by the next larger earthquake. In situations where there is no next larger earthquake, the open interval is used. The ETAS+ likelihood is then maximized only over these complete intervals. No additional parameters are required. I here show the impact of positive statistics on forecasting, applying ETAS+ to simulated and real aftershock sequences. For each sequence, I generate probabilistic forecasts using both classical techniques and ETAS+ and compare the forecast success. There is no difference in the way stochastic catalogs are generated for classical and ETAS+ parameters. ETAS+ improves the forecast relative to a uniform completeness model, without the need for a detection model or additional parameters.

Key Words
forecasting, ETAS, catalog completeness

Citation
van der Elst, N. J. (2024, 09). Using ETAS+ to forecast aftershock sequences. Poster Presentation at 2024 SCEC Annual Meeting.

Related Projects & Working Groups
Earthquake Forecasting and Predictability (EFP)