Tsunami-like occurence detection and characterization: a comparison between time-frequency analysis methods

Tsunami-like waves having periods of about 30 minutes are often detected in the Gulf of Genoa and account for a relevant part of the coastal sea-level variability. In some of the cases, their amplitude is comparable or even higher than the local tides. Eight years of sea-level data were analyzed with the aim of assessing their occurrence, intensity and generation mechanism. Data were sorted on annual basis to infer about possible seasonality. Spectral characterization in the "tsunami-band" shows a bimodal oscillation centered on 26.7 and 30.1 minute period common to each year sub-set. A timefrequency analysis based on FFT allowed identifying the occurrences of the most relevant events, while wavelet analysis demonstrated to be a more appropriate tool for the characterization of those which presented relevant differences in terms of intensity, duration, time-frequency response.