Persistence Analysis Validation

The Kuwashima and Hogben method [6] for the estimation of persistence of storms and calms was based on that originally developed by Graham [34]. Taking Graham's method as a starting point, the approach taken by Kuwashima and Hogben was first to examine its validity by applying the method to a selection of cases in which extensive measured data sequences were available, and then to modify the method to improve its reliability.

The modified method was then tested by applying it to a range of cases where persistence curves derived from measured data sequences (but not the actual data sequences) were available, and further improvements were made.

The method thus evolved, and was shown to give predictions of persistence statistics for both exceedance and non-exceedance of wave height which were in good agreement with measured data at a wide range of sites. The sites were mostly in UK waters, but also included two sites remote from the UK (one in Hong Kong and one in South Africa). The agreement with this range of sites offers reassurance that the method, though empirical, has wide validity.

Reference [6] gives details of comparisons of the method with measured data at 7 different sites. For the purposes of this section it is sufficient to present just one comparison. The data for Slangkop Point (in South Africa) is compared with the method in Figure 20.

Figure 27 - Slangkop Point - Persistence of Storms Figure 28 - Slangkop Point - Persistence of Calms

Figure 27 shows the persistence of storms plotted in terms of number of occurrences experienced over a period of about 6 years (Feb 1976 - Aug 1982). The plot is presented on log axes, which is similar to one of the graphical presentation options in Global Wave Statistics Online. It can be seen that agreement is remarkably good over a range of significant wave height thresholds from 1m to 6m.

Figure 28 shows the equivalent plot for the number of calms experienced over the same period. Again the agreement is excellent.

(Note that this comparison has been made using the NMIMET program, and using a visual dataset specially localised on the Slangkop location. If you attempt to repeat this validation using Global Wave Statistics Online the result will be somewhat different. Although the Slangkop location (3407'S 1810'E) is in Sea Area 85, this covers quite a large area and furthermore the location is quite close to the boundary with Area 90.)

It is obviously preferable to derive persistence statistics from long continuous records of wave height, but such records are seldom available for the site of interest. When time series data is not available, the Kuwashima and Hogben method offers predictions of persistence statistics from the probability distribution of wave height which have been shown to agree well with measured data for both exceedance and non-exceedance at a number of sites. In particular the method has been shown to give reasonably reliable results when used with the 3-parameter Weibull coefficients fitted to visual wave data by the NMIMET program and contained in the Global Wave Statistics Online database.