Detecting tsunamigenesis from undersea earthquake signals

Abstract

Tsunami warnings have always been made from imperfect and limited data often relying solely on the earthquake magnitude and epicenter location. Large earthquakes (M > 7) occurring in a deep sea, often cause significant tsunamis. However, tsunami generation is not determined by the magnitude of the earthquake alone, but also affected by the fault rupture mechanism and type of fault. Thus, the prediction of tsunami generation is very complex. Since the earthquake signal travels 20-30 times faster than tsunami, the seismic signals can be used for tsunami prediction and warning. For accurate prediction of the generation of a tsunami, various parameters of the tsunami source have to be obtained in near real-time. Besides the earthquake magnitude and the epicenter location, the fault rupture area, the amount of slip and the orientations of the fault plane are important indicators for tsunamigenesis. This study aims to study the computation of these parameters in near real-time (i.e. few to tens of minutes after an earthquake of M > 7 happened in deep sea). These parameters will be used to run tsunami propagation modeling to compute tsunami arrival times, wave heights and run-up at the shores. This tsunami prediction was further enhanced by signals analysis methods, namely Fast Fourier Transform (FFT) and Continuous Wavelet Transform (CWT). These analyses were performed on the time series records of both the tsunamigenic and non-tsunamigenic earthquakes in the South East Asia and the South China Sea region. The FFT revealed that tsunamigenic earthquakes are rich in long-period energy and depleted in high-frequency energy. The CWT not only confirmed this finding but also showed the time when these frequencies occurred in the signal. The long-period energy is mainly associated with slip on the fault plane. Therefore, the proposed methods of FFT and CWT can successfully extract the slow rupture and large slip phenomena from the seismic signals of tsunamigenic earthquakes. This finding can be used for diagnosis of tsunamigenesis in the advancement of timely tsunami prediction and warning. © 2009 Elsevier Ltd. All rights reserved.