Aftershock forecasts based on incomplete earthquake catalogs: ETASI model application to the 2023 SE Türkiye earthquake sequence

Abstract

The Epidemic-Type Aftershock Sequence (ETAS) model is the state-of-the-art approach for modeling short-term earthquake clustering and is preferable for short-term aftershock forecasting. However, due to the large variability of different earthquake sequences, the model parameters must be adjusted to the local seismicity for accurate forecasting. Such an adjustment based on the first aftershocks is hampered by the incompleteness of earthquake catalogs after a mainshock, which can be explained by a blind period of the seismic networks after each earthquake, during which smaller events with lower magnitudes cannot be detected. Assuming a constant blind time, direct relationships based only on this additional parameter can be established between the actual seismicity rate and magnitude distributions and those that can be detected. The ETAS-Incomplete (ETASI) model uses these relationships to estimate the true ETAS parameters and the catalog incompleteness jointly. In this study, we apply the ETASI model to the SE Türkiye earthquake sequence, consisting of a doublet of M7.7 and M7.6 earthquakes that occurred within less than half a day of each other on February 6, 2023. We show that the ETASI model can explain the catalog incompleteness and fits the observed earthquake numbers and magnitudes well. A pseudo-prospective forecasting experiment shows that the daily number of detectable m ≥ 2 can be well predicted based on minimal and incomplete information from early aftershocks. However, the maximum magnitude (Mmax ) of the next day’s aftershocks would have been overestimated due to the highly variable b value within the sequence. Instead, using the regional b value estimated for 2000-2022 would have well predicted the observed Mmax  values.