Time-frequency and time-scale analysis applied to soliton detection in the Caribbean sea

Abstract

Time-frequency and time-scale analysis methods are used to detect solitons, south of Puerto Rico, in the Caribbean Sea. A string of underwater sensors, deployed at particular depths, produced various temperature time series currently being analyzed for soliton detection. Different causes, such as a non-optimal placement of the sensor array in time or space, could yield a weak soliton signature deeply buried within the signal. Underwater sensor deployment is expensive, thus, the analysis methods were calibrated by incorporating a synthetic soliton into the original signal. Its parameters, such as peak amplitude, peak to peak distance and number of peaks, among others, were varied in order to "ground truth" the analysis results. The short time Fourier transform (STFT) and wavelet transform (WT) were used to analyze the data. The combined assessment of results, obtained through different methods, were used to pinpoint possible solitons in the time domain.