matloa is available In two dimensions, such as a SASS or 

 SEABEAM survey area, it might be possible to verify through 

 regression analysis a simple 6 dependent function to describe 

 a and b. These functional forms, fg(9)=a and f^(9)"b, could 

 then define the best model for estimating model coefficients 

 in areas where only randomly oriented track data is available. 



(5) Extension of the functional representations beyond the spatial 

 frequency at which the surface sonar encountered its white- 

 noise level, would be attempted by studying the spectra of 

 deep-towed sonar and bottom photographs. It is anticipated 

 that a general functional form, perhaps a simple extrapolation 

 of the power law form, will be discovered. In the many areas 

 where high resolution bathymetry is not available, this 

 extrapolated function should provide the best available esti- 

 mate of spatial frequency stmicture at very short wavelengths. 



(6) Incorporation of the model into existing data sets would be 

 the final development stage. This roughness model would be 

 calculated for 5' grid cells and merged with the existing 

 gridded data sets being developed at the Naval Oceanographic 

 Office. All grid cells within a homogeneous province would be 

 represented by coefficients generated from data located any- 

 where within that province. By Integrating this stochastic 

 model with the existing deterministic models of oceanic 

 depths, an essentially complete description of the sea floor 

 will be contained in a single data base. This data base, when 

 combined with similar gridded models of sea water sound veloc- 

 ity, sediment column sound velocity, sediment thickness, and 



43 



