In many scientific endeavors of a computational nature, the representation or the simulation of natural phenomena requires the use of grids. As a matter of fact, such processes are often described by complex sets of equations like partial differential equations that rarely have simple analytical solutions when it becomes necessary to mimic reality. In particular, in geophysics, 2-D or 3-D grids are commonly used for such purposes as finite-difference schemes applied to wave-propagation analysis or traveltime computation.

GOCAD is a software for handling surfaces in 3-D and therefore allows the
representation and the definition of sophisticated 3-D models. The definition
of 3-D grids is possible in GOCAD via the `GRID3_t` object Mallet (1993),
but once the empty grid has been allocated, it is necessary to assign to each
node of the grid a corresponding value of the variables extracted from the
model.

In this article I try to verify the feasibility of designing such a grid
especially for complex models, and then I propose a new implementation of
the `GRID3_t` object in a manner more convenient for geophysical and
geological purposes.

Figure 1

% real V ; /* Velocity */ % real d0 ; /* Density */% function density() ; % function K_compute() ;

Domain_3 : S1-, F-, S2+ ; /* Portlandian */ { V = 3100.0 ; d0 = 2.75 ; }

Domain_4 : S2-, F-, S3+ ; /* Kimeridgian : clay */ { V = 2700.0 ;

density( input1, output1 ) ;

real porosity = 0.37 ; }

Domain_5 : S3-, S4+ ; /* Oxfordian : salt */ { V = 4200.0 ;

real K0_x = 3300 ; /* Porosity */ real K0_y = 456 ;

K_compute( point, result ) ; }

11/17/1997