With a one-dimensional array, even with sensors in per- 

 fectly known positions, we obviously cannot fully determine 

 the modal distribution of the waves present. Other information 

 is required, and this is obtained from knowledge of the mean 

 vertical structure. Knowing both o" and k, we can obtain the 

 x-phase speed as ^ 



^X '= }<, " (Xi ' Xj) sin Q 



where c is the wave-phase speed and © is the angle between the 

 wave number vector and the normal to the array axis ( see Fig. 2-1). 

 Clearly c will always be greater than or equal to c , Using 

 theoretical knowledge of c (from the computer program written for 

 this purpose) it is possible to deduce the modal structure for a small 

 number of modes. 



We are helped in the determination of modes by relationship (1) 

 which says that at a given frequency, the modal wave-lengths will 

 differ by ratios of integers. Thus if mode 1 has a wavelength of 

 2 km. (which is true at about 14 minute periods) then mode 2 will 

 have a wavelength of only 1 km., and a correspondingly smaller 

 phase speed. 



With 3 sensors we can perhaps distinguish 2 modes at any given 

 frequency. If, however, a few modes are dominant, we can make 

 use of the so-called "group-delay" 



which is related to the theoretical group velocity, and gives additional 

 information from adjoining frequencies. 



In the Bermuda area at a depth of 700 m., a 1 degree temperature 

 change corresponds to 100 m. of water displacement. We have set 

 a maximum permissible vertical excursion of 3 m. (.03 degree) or 

 about 3 times the stipulated sensor sensitivity. The pressure trans- 

 ducers will detect vertical motion greater than 1.5 m. which will 

 permit us to partially compensate the temperature records. 



We will continue to rely on the availability of Panulirus station data 

 to provide sufficient information about the vertical to compute the 

 normal mode decomposition. 



There are many difficulties involved in this project, and the utility 

 of these methods and the hardware design have to be proven. The 



199 



