This paper deals with the effect of the maximum allowed error (MAE) on the segmentation of significant wave height time series using the bottom up approach. This approach was implemented in Soukissian, Samalekos (2006), who adopted a constant MAE in order to identify and analyze the developing, decaying and stationary -sea states and their associated durations. The MAE corresponds to the representation error (obtained from the method of least squares) and determines the length of each segment (i.e., the sea state duration) and the overall quality of the linear fit. Different values of the MAE provide different populations of developing, decaying and stationary sea states and associated durations. In this paper a detailed sensitivity analysis of the effect of the MAE on the duration and intensity statistics of the obtained sea states is made.
The analysis of large amount of wave data accumulated over long-time intervals (as e.g., some decades) is a critical task in extracting useful information about the variability of wave conditions at a particular site. Duration statistics of sea states extracted from historical Hs data can provide important information for marine operations and can be used as a complementary tool in operational wave forecasting, wave climate studies and extreme wave analysis. In the relevant scientific literature two types of sea state duration analysis can be found:
Duration of stationary sea states and
duration (persistence) of storms.
In the first case, Hs -time series is treated as a piecewise stationary process; see e.g., Labeyrie (1990), Athanassoulis and Soukissian (1991), Athanassoulis, Vranas and Soukissian (1992), Soukissian and Theochari (2001). In the second case, the persistence (duration) of sea states above a given threshold level is considered; see e.g. Graham (1982), Mathiesen (1994) Anastasiou, Tsekos (1996), Tsekos, Anastasiou (1996), Sobey, Orloff (1999) and Jenkins (2001).
