according to the nearest point on the JNWP grid, and the wind fields at 

 these times were translated to the point half way between and averaged in- 

 side a circle of appropriate radius. After all lows were treated in this 

 way, points not translated were simply averaged. The procedure treated 

 the moving low pressure systems in such a way as not to distort the wind 

 field around the lows. The sub -tropical highs could be averaged, or inter- 

 polated, without translations. 



The Wave Hindcasting Procedure 



Once adequate representations of the wind fields were available 

 every six hours, it was possible to generate the wave hindcasts according 

 to the application and extension of the results of Baer (1962), Moskowitz 

 (1964), Pierson and Moskowitz (1964) and Pierson (1964). The hindcasting 

 procedure accounts for the effects of spectral growth (both due to durati'~'n 

 and fetch), propagation, and the dissipation by turbulence due to a local 

 wind sea of those spectral components that are traveling against the wind. 



RESULTS 



In this paper, onJ.y a very small portion of the results can be given. 

 For each point in the field a set of 180 numbers was obtained to describe 

 the wave spectrum every six hours. The numbers represent the contribu- 

 tion to the total variance within given ranges of frequencies and directions 

 in the following representation. 



^2 r-^2 



/ 



f. 



/ S(f. 0)d0 



^ 



df 



Fifteen frequency ranges and twelve direction ranges for a total of 180 

 values were obtained. 



A sample of one such spectrum for 17 December 1959 at 06Z at 

 grid point 72 is given in Table 1. The units are (ft)^ . The direction and 

 frequency ranges are shown. Also tabulated is the sum over all directions 

 for a given frequency range. 



One of the developmental samples was from about 15 December 1959 

 to 2 8 December 1959- A graph of the significant wave height as observed 

 and as hindcasted at four surrounding grid points is shown in Figure 1. 

 The waves range in significant height from 12 feet to nearly 40 feet and the 

 hindcasting procedure tracks the hourly variation in wave height quite well. 



Figure 2 shows a scatter diagram of observed versus predicted heights. 

 The observed heights are instrumentally observed and have been com- 

 puted in terms of the area under a frequency spectrum (corrected for in- 

 strument response) estimated from a 15 minute long wave record. The 

 sampling variability is about 10% of the value of the observed height and 



75 



