with either U2 or F2 whichever comes first from the left side of the 

 figure. If U comes first, Z/2 is added to the duration at this point, 

 and the U2 ordinate is followed to either this new duration or to the 

 F2 whichever is first from the left side of the chart. (Compare with the 

 preceding paragraph.) At this point, Hp2' ^F2' hn2> ^^^ ^m2 inclusive, 

 are read off. If the constant energy line had intersected F2 before U2, 

 it is only necessary to drop down along the F2 abscissa to its intersec- 

 tion ^'ith U2, and at this point read H^2> '^F2> ^2 ^^^ ^m2- (This pro- 

 cedure could be used for many cases in which U2 is greater than Uj.) 



The major differences in technique are used when U2 is less than Ui 

 and the H^T^ = constant line from the intersection of Ui and tmi + Z/2 

 does not intersect either U^ or F . Forecasting theory used here pre- 

 dicts that waves due to a constant wind blowing over an unlimited fetch 

 for an unlimited duration will eventually reach limiting height and period 

 distributions beyond which growth will not continue. In Figure 3-16 the 

 limit of this state is delineated by the line labelled maximum condition. 

 To the right of this line, it is assumed that any energy transport to the 

 waves by the wind is compensated by wave breaking, hence no wave growth 

 occurs . 



3.52 EFFECTS OF MOVING STORMS AND A VARIABLE WIND SPEED AND DIRECTION 



In principle, it should be possible to extend the Inoue differential 

 equation for wave growth to highly irregular conditions, but no experi- 

 mental verification of this concept has been published. Kaplan (1953) 

 and Wilson (1955) have proposed techniques for applying the simplified 

 prediction techniques to variable wind fields and changing fetches. The 

 procedures appear reasonable and these techniques are used, although no 

 statistics are available for verification. 



3.53 VERIFICATION OF SIMPLIFIED WAVE HINDCAST PROCEDURES 



Comparisons of hindcast wave heights and observed wave heights, 

 similar to Figure 3-7, have been given by Jacobs (1965) for the PNJ wave- 

 prediction system, by Bates (1949) and Isaacs and Saville (1949) for the 

 early Sverdrup-Munk method, by Kaplan and Saville (1954), for the early 

 SMB method, and by Bretschneider (1965) for a later revision. The basic 

 data from which the prediction curves were derived, siommarized by 

 Bretschneider (1951), also indicate the range of variation that may be 

 expected. 



It is generally believed that much of the discrepancy between observed 

 and predicted waves is random, and that statistical summaries of observa- 

 tions and predicted values will agree much better than the individual 

 observations. Saville (1954) and Pierson, Neumann and James (1955) give 

 summaries of results from a systematic program for deepwater hindcasting 

 waves at the four locations shown in Figure 3-17. The U.S. Naval Weather 



3-40 



