Tsunami detection methods for ocean-bottom pressure gauges

Abstract

The real-time detection of tsunami waves is a fundamental part of tsunami early warning and alert systems. Several algorithms have been proposed in the literature for that. Three of them and a newly developed one, based on the fast iterative filtering (FIF) technique, are applied here to a large number of records from the Deep-ocean Assessment and Reporting of Tsunamis (DART) monitoring network in the Pacific Ocean. The techniques are compared in terms of earthquake and tsunami event-detection capabilities and statistical properties of the detection curves. The classical Mofjeld's algorithm is very efficient in detecting seismic waves and tsunamis, but it does not always characterize the tsunami waveform correctly. Other techniques, based on empirical orthogonal functions and cascade of filters, show better results in wave characterization but they usually have larger residuals than Mofjeld's. The FIF-based detection method shows promising results in terms of detection rates of tsunami events, filtering of seismic waves, and characterization of wave amplitude and period. The technique is a good candidate for monitoring networks and in data assimilation applications for real-time tsunami forecasts.