Application of Formula

Plotting Position Formula

Traditional

^

(Gumbel 1958)

Fisher-Tippett I (FT-I)

^

(Gringorten 1963)

0.27

^

Weibull Distribution Function

0.23

(Goda 1988)

0.52

Fisher Tippett II (FT-II)

^

or Frechet Distribution Function

0.11

(Goda and Kobune 1990)

Log-Normal Distribution Function

^

(Blom 1958)

Parameter definitions:

1

For censored data, *N *should represent the total number of events over the time interval considered

(not just the number of censored events)

prudent to plot the computed distribution function and data together and ensure that the fit is consistent with

good engineering judgement.

(4) Outliers. Outliers are retained in the data, but they should receive special scrutiny, as follows:

(a) Ensure accuracy. Each outlier should be checked to ensure that it is a valid data value, rather than

a measurement or modeling error.

(b) Examine each event that produces a high outlier. Typical causes are very severe winter storms or

direct impact of an intense hurricane. If extreme events at the site are produced by distinctly different natural

processes (different statistical *populations*), it may be preferable to divide the data values into several series,

one for each process, and analyze each series separately (e.g. Goda 1988). For example, winter storms and

hurricanes should not necessarily be expected to produce extremes that follow the same extremal probability

distribution function. Extreme data values can be analyzed as separate populations only if sufficient data

values are available in *each *population.

Hydrodynamic Analysis and Design Conditions

II-8-11

Integrated Publishing, Inc. |