Another way is to take wave records from a moving ship by means of 

 the shipborne wave recorder as described by Cartwright [1956]. The ship is 

 run on courses corresponding to an n sided polygon and the shift in frequency 

 of the spectral components is studied. With enough degrees of freedom per 

 spectral estimate, it should be possible (in principle) to resolve the spectral 

 estimates into a directional spectrum by the inversion of some simultaneous 

 linear equations in an appropriate number of unknowns somewhat along the 

 lines of the method described by Pierson[l952]. However, if the response of 

 the instrument to the waves is different for different headings due to the pres- 

 ence of the ship and if the records are too short so that sampling variation 

 from record to record is pronounced, then the difficulties to be encountered will 

 be even greater than those encountered in this report. Although some of the 

 data reduction might be eliminated by analogue methods, the procedure would 

 have essentially the same degree of complexity as the one used in this report. 



The latest proposed procedure for determining directional spectra is 

 given by Longuet-Higgins [1957], The records from an airborne altimeter 

 capable of measuring •n(x,y) and 9T](x,y)/9t are assumed at the starting point, 

 and then by computing various moments from the data as determined by such 

 quantities as the average distance between successive zeros at various head- 

 ings and the velocity distribution of zeros, the moments of the spectrum are 

 obtained^ Then by an inversion technique the spectrum is deduced, 



Pierson[l952] proposed the use of anairborne altimeter to determine 

 the directional spectrum. The method of analysis involved the study of the 



247 



