obtained with a deterministic simulation model must be viewed as a representa- 

 tive result that has smoothed over a large number of unknown and highly 

 variable conditions. 



86. Similarly, in use of a deterministic model in a predictive mode, 

 the factors responsible for beach change (in the case of GENESIS, primarily 

 the waves) are not known. A time series of wave height, period, and direction 

 must be forecast for use in the prediction and can be considered as only one 

 of many possible wave climates that might occur. 



Accounting for variability 



87. Since there is great variability in the nearshore system, any one 

 prediction of shoreline change cannot be the correct answer. Several studies 

 have been made on wave variability and shoreline change prediction (Kraus and 

 Harikai 1983; Le Mehaute, Wang, and Lu 1983; Kraus, Hanson, and Harikai 1984; 

 Hanson and Kraus 1986a; Hanson 1987; Walton, Liu, and Hands 1988), and some 

 guidance has been developed for use in the prediction process. These referen- 

 ces should be consulted to supplement discussion given here. 



88. A simple procedure used at CERC to estimate the effect of wave 

 variability is to compute the standard deviation of the wave height and 

 direction in the input wave time series and then adjust values of the input 

 waves through a range defined by these deviations. GENESIS allows adjustment 

 of wave height and direction by user -specified amounts. Wave period is not 

 normally varied, but in certain applications, such as a situation involving 

 waves of long periods or a sea bottom with highly irregular features, the 

 refraction pattern will be particularly sensitive to wave period. Another 

 procedure uses different hindcast time series if such data are available. By 

 varying the input wave height and direction within a physically reasonable 

 range, a series of shoreline change predictions is made within which the 

 actual change is expected to lie. Variation of input parameters is also part 

 of the sensitivity analysis to be performed to obtain some idea of model 

 dependence on empirical parameters, as discussed in a later section. 



43 



