Empirical Low‐Frequency Earthquakes Synthesized From Tectonic Tremor Records
Satoshi Ide
Abstract
Abstract Tectonic tremor and low‐frequency earthquakes (LFEs) are two different representations of the high‐frequency (>1 Hz) components of broadband slow earthquakes, which have been discovered in various tectonic regions. Although LFEs are considered building blocks of tremor, it is difficult to find constituent LFEs for some long‐duration tremor sequences. Here we introduce a new scheme to synthesize an impulsive response from a tremor source, or empirical LFE (eLFE), from complex tremor sequences. The eLFE is constructed by maximizing the absolute amplitude ( L ∞ norm) of the normalized and stacked waveforms of many tremor sequences at many stations. This problem includes a highly multimodal objective function, which is maximized via a combined optimization method that incorporates a genetic algorithm and iterative improvements. We first apply this method to several validation tests using synthetic tremor waveforms and real LFEs, and then tremor activities where few nearby LFEs have been detected. One example is the Okayama tremor, which has been known to be extraordinarily sensitive to tidal stress. The eLFE is helpful in determining the accurate hypocentral location and recovering the long‐term comprehensive tremor activity via a matched‐filter analysis. The well‐constrained depth of the Okayama tremor sources is less than 30 km, which suggests that the subducting Philippine Sea Plate is very shallow at this distant location from the trench axis. This eLFE scheme highlights the fact that tremor sequences may be composed of random fluctuations that are not necessarily impulsive and emphasizes the importance of characterizing such randomness to better understand slow earthquake behavior.