J. Rossouw


Design wave heights were estimated from measured data using a virtually continuous data set consisting of 8 years of waverider data. Evidence is given which shows that waveriders tend to malfunction during storms. Special care was taken to select independant and identically distributed samples from the data before fitting a number of probability distributions to the selected wave heights. Bootstrap techniques were used to select the models that give the best fit to the data as well as to determine the confidence bands for the predicted design wave heights. It is shown that once the model for the long term distribution of wave height is chosen, relatively narrow confidence bands can be obtained for the most probable value of up to the 100 year return period wave if maximum use is made of the available data. Uncertainty about the selection of the model and the representativeness of the measurements however reduces the usefulness of these confidence bands. A plea is also made in the paper to stop using the concept of a wave with a certain recurrence interval but rather to specify a wave with a given risk of being exceeded within the design life of the structure.


design wave; probability density functions

Full Text: PDF

Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.