to a smaller grid cell spacing Is the storage requirements of the data. 

 For example, the 5' grldded data set extended from TO'S to yO'N being 

 developed by the Naval Oceanographlc Office will require approximately 

 five million storage locations. To extend this grid to the order of 100 

 meters spacing (assuming this was determinable globally) would require 

 approximately 5 x iqIO storage locations. 



In light of the physical and practical limits of describing high 

 spatial frequency sea-floor topography determinlstically, the apparent 

 alternative is some probabilistic (stochastic) model describing the dis- 

 tribution of features in a statistical sense rather than determining the 

 exact locations of depths. To be useful, this stochastic model must 

 describe, with a reasonable number of parameters, most of the true vari- 

 ability of a region of sea floor. Using a stochastic representation of 

 high frequency feature distribution in combination with the lower fre- 

 quency deterministic models being developed, an essentially complete 

 description of the sea floor is possible. The development of such a 

 stochastic model of sea-floor topography is the intent of this 

 dissertation. 



