The UNIMAK data and the VALLEY FORGE data do not indicate divergent trends, 

 that is, the methods 6,7 of wave estimation used by the observers on the UNIMAK and the 

 VALLEY FORGE give approximately the same characteristic wave height. 



The U.S. Weather Bureau data, on which the long-term distributions given in this re- 

 port are based, comprise between 11,000 and 18,000 separate observations for each ocean 

 station. It is concluded that the errors associated with the visual observations are fairly 

 well averaged out when such a large number of observations are utilized to define the distri- 

 bution and that the reported characteristic wave heights are therefore proportional to the 

 severity of the sea. 



Further evidence to support the validity of the Weather Bureau data can be drawn from 

 an analysis of measurements of wave height recently made by J. Darbyshire 4 by means of a 

 wave meter installed on a weather ship. These measurements were made over a period of 

 about one year, February 1953 to January 1954, at North Atlantic Weather Stations I and J; 

 see Figure 8. Darbyshire reported the maximum wave height for each 3-hr period for which 

 visual wave observations were made while the ship was at sea. The visual observations 

 made by weather observers are reported as the "characteristic" wave height. According to 

 Appendix B the characteristic height is proportional to the significant height. It is of interest 

 to compare the visual observations with the measurements obtained with the wave meter. If 

 the hypothesis is accepted that the short-term distribution of wave height follows the Rayleigh 

 distribution, then the maximum significant and characteristic wave height for any given sea 

 condition are related by a constant factor. Thus the long-term distributions of maximum and 

 characteristic wave heights should be of the same type, log-normal in this case, and should 

 differ only in their mean values. The U.S. Weather Bureau data indicates that the standard 

 deviation* of log e (characteristic wave height) is 0.622 at Station J and 0.612 at Station I 

 as compared with a value of 0.57 for log e (maximum wave height) for the measurements at 

 Stations I and J reported by Darbyshire; see Figure 8. A log-normal distribution has been 

 fitted to the wave-meter data on the assumption that the distribution of maximum wave heights 

 is log-normal. The experimental data indicate excellent agreement with the fitted distribution, 

 well within the accuracy of the measurements. The latter fact, together with the good agree- 

 ment between the standard deviations of characteristic (visual estimates) and maximum (meas- 

 urements) wave heights, supports the hypothesis that the distribution of wave heights may be 

 approximated by Rayleigh and log-normal distributions for the short and long term, respective- 

 ly. In a recent article 5 Darbyshire tests the applicability of the long-term log-normal distribu- 

 tion to extensive data on maximum wave heights obtained by use of the British wave meter. 

 He concludes that the logarithmic law appears to be a useful guide to determine the incidence 

 of a particular wave state at a given location. 



*The numerical values given here apply for wave heights measured in feet 



13 



