shoreline change occurring over several years, unusually high wave energy and 
low wave energy years in the hindcast were avoided. The consequences of 
severe storm events are treated with the beach erosion model, discussed in 
Part V. 
43. The representative wave data set is to be used for both calibration 
and verification of historical (measured) shoreline change and prediction of 
future shoreline change. Since wave data are not available in either case, a 
deterministic interpretation of the input wave data set is not possible. 
However, one can obtain a representative data set with which certain para- 
meters can be varied to estimate the range of variability given by the model 
(the wave and shoreline models). To keep computer file sizes within manage- 
able limits, 3 years of data were selected from among the 20 years of avail- 
able hindcast data to produce the representative wave data set. 
44, From consideration of Table 1, the consecutive years 1973 through 
1975 were judged to be suitable to represent the total data set. The average 
significant wave height for these 3 years is equal to the average wave height 
for the 20-year total. One year, 1973, has an average significant wave height 
above the 20-year average, and each of the selected 3 years contains two of 
the highest ten wave energy events occurring among the 20 years. The average 
peak spectral period for these years ranges from 6-8 sec, which gives a spread 
of values around the commonly occurring 7-sec peak period. Since wave refrac- 
tion and shoaling are to a large extent controlled by wave period, this spread 
is a desirable feature. 
45. As mentioned above, the chosen 3-year wave set possesses an average 
wave height equal to the average of the available 20 years of hindcast data, a 
characteristic compatible with the function of the shoreline change model. A 
complementary characteristic is that the data set includes slightly more high- 
wave energy events than average to account for realistic stormier conditions. 
It is noted that the 3-year record employed is much longer than records typi- 
cally reported in shoreline modeling literature, in which only a small number 
of representative conditions, such as seasonal averages, are repeated numerous 
times over the simulation period. 
30 
