given in Table 11 are intended for comparison with the hindcast values but not 

 for use in constructing long-term extreme wave conditions. It is a generally 

 accepted rule of thumb that any extrapolation beyond twice the extent of the 

 data becomes highly unreliable (Borgman and Resio 1982); hence, 2 years of 

 data may be useful for predicting 2 to 5 years of wave conditions but are not 

 very reliable for 5- to 100-year predictions. 



Discussion 



24. The use of measured data to make long-term wave condition estimates 

 has been discussed in several recent articles (Carter and Challenor 1978, 

 Challenor 1982, and Wang and LeMehaute 1983). The key aspect of the theory of 

 extremes is that in looking at enough years of data, the chance of seeing most 

 types of hazardous events is good. In this respect, measured data will become 

 useful as longer records are made available. At this time, hindcast data ful- 

 fill the need. The wave data that were measured at Kodiak provide an inter- 

 esting comparison to hindcast results. 



25. It is surprising that the outer buoy, or buoy 1, did not display at 

 least one maximum significant wave height in exceedance of 6 m. If it is as- 

 sumed that the hindcast grid point represents the same general wave condition 

 as the outer buoy, and since there were 78 events producing significant wave 

 heights of 6 m or more in the 20-year hindcast time period, the probability 



of seeing two consecutive years with no significant wave heights greater than 

 6 m at the outer buoy is nearly zero. Because of this, it is not possible to 

 either support or contradict the results from the hindcast analysis using the 

 measured wave data. If consecutive hindcast and measured values were avail- 

 able, this behavior could be studied; however, such is not the case at this 

 time. 



26 



