A selected record of the measured wind wave data, which contained the maximum wave height during a mistral event, was divided into five segments, and computations of wavelet bicoherence were conducted for the whole record and for all divided segments in order to analyze the details of nonlinear wave-wave interactions.

In computing wavelet bicoherence, both sum and difference rules of frequencies were considered. A close inspection of the computed wavelet bicoherence for the whole record of wave height and the divided segments reveals that the details of wave-wave interactions are better explored by considering five-minute segments of wave height. Averaging over the whole record of wave height causes loss of important temporal characteristics in nonlinear wave-wave interactions.