5.7: The 100-Year Wave - Estimate from Short-Term Statistics
- Page ID
- 47250
For long-term installations, it is important to characterize the largest wave to be expected in an extremely large number of cycles. We will make such a calculation here, although the Rayleigh distribution does not adequately capture extreme events over such time scales. Spectra and the consequent Rayleigh height distribution are short-term properties only.
The idea here is to equate \( p(h > h_o) \) from the distribution with the definition that in fact \(h > h_o\) once in 100 years. Namely, we have
\[ p (h > h_{100 yr}) \, = \, \dfrac{1}{100 \textrm{ years} / \bar{T}} \, = \, e^{-2 h_{100 yr}^2 / \bar{h}^{1/3}}, \]
where \(\bar{T}\) is the average period. As we will see, uncertainty about what the proper \(\bar{T}\) is has little effect in the answer. Looking at the first equality, and setting \(\bar{T} = 8\) seconds and \(\bar{h}^{1/3} = 2\) meters as example values, leads to
\begin{align*} 2.5 \times 10^{-9} \, &= \, e^{-2 h_{100 yr}^2/ \bar{h}^{1/3}}; \\[4pt] \log (2.5 \times 10^{-9}) \, &= \, -2 h_{100 yr}^2 / 4; \\[4pt] h_{100 yr} \, &= \, 6.3 \textrm{ meters, or } 3.1 \bar{h}^{1/3}. \end{align*}
According to this calculation, the 100-year wave height is approximately three times the significant wave height. Because \(\bar{T}\) appears inside the logarithm, a twofold error in \(\bar{T}\) changes the extreme height estimate only by a few percent.