An Automatic Normal and Abnormal Wave Events Classification Algorithm: Using Continuous Wave Monitoring Data at a Fixed Station

Abstract

A classification algorithm for the time-series wave data into normal and abnormal wave periods was proposed and applied. Unlike the existing traditional wave duration (persistence) analysis methods, this method divides wave height data into independent individual wave events with objective and automated criteria, so using various criteria and analyzing sensitivity is possible. This technique for detecting peak wave heights and determining the influencing period of the wave heights was applied to the KMA (Korea Meteorological Administration) wave height monitoring time series data, which is a typical type of marine environment observation data. As a result of the application, it was found that a more stable and appropriate wave event periods determination classification is possible when smoothing short-term wave height fluctuations. In addition, it was found that rather than detecting peak waves, it was found to have a sensitive effect on the almost wave-based quantitative criteria that determine the time of growth and decay of these high waves. On the other hand, the statistical characteristics of abnormal and normal wave events were found to show significant differences in mean, variance, and temporal change patterns.