Diffusion problem

[Jim Westervelt]

Regarding the visualization of an underground blast: The best GRASS has
to offer is probably r.mapcalc.  This program provides a cellular
automaton approach in that users write instructions/code that describes
how one or more grass layers are mathematically and logically combined
into some result.  Instructions can be written that utilize information
contained in neighboring cells.  These instructions are then applied to
each individual grid cell in the computation of output captured in a
single map.   Difficulties will arise with GRASS's inability to store
floating point numbers (though r.mapcalc performs all computations in
floating point) and the limitation on r.mapcalc to produce only a
single resultant map.    You might want to contact
shapiro at zorro.cecer.army.mil regarding a copy of a lengthy tutorial with 
examples for r.mapcalc.   Hope you have access to a very good number
cruncher!

Something that may be available in the future to help with hydrologic
modeling is a finite-element analysis program for raster data.