The limitations of statistical low runs prediction in rough seas: a study based on real wave data

This work aims to estimate the feasibility of wave runs statistical prediction in rough seas, with a specific application of Kimura’s theory (Kimura, A.,1980). According to Kimura, the probability of a low run of waves of assigned length is a function of the correlation between consecutive waves in a group. Time series from wave buoy measurements (located at Bideford Bay, England and data downloaded from the Channel Coast website) have been tested. The spectral analysis showed that higher values of the significant wave height correspond to broader bandwidth spectra. The Kimura’s low runs prediction is noticeably affected by this aspect: by plotting the low runs probability against the number of incoming waves, narrow spectra show a steeper behaviour of the low runs probability curve, whereas broader spectra correspond to flat probability curves settled on lower values. Therefore, it appears that broader spectra are characterised by more grouping phenomena. Moreover, the Kimura’s estimation has been calculated in an adaptive frame applied to the signal, in order to test the sensitivity to the length of the data; a minimum length of 6 to 8 hours has been found to have a stable prediction of the low runs probability