How to implement seismic or any other type of ray-based tomography in
a new class of cellular parameterization built from any chosen distrib
ution of nodes in two or three dimensions is shown. The complete flexi
bility of this novel approach allows one to introduce detail in a tomo
graphic model only where desired, thereby reducing memory and computat
ion time, or to impose complex a priori constraints on the inversion.
Full details of powerful new algorithms are given to generate unequall
y sized tetrahedral or polyhedral cells and to calculate the necessary
Frechet derivatives required in linearized tomography. These algorith
ms are efficient enough in three dimensions, to allow the parameteriza
tion to be refined during an inversion. The methods are illustrated wi
th numerical examples. It is concluded that in linear or nonlinear inv
ersion the computational cost of the new algorithms will not be signif
icantly higher than that incurred by using a regular Cartesian grid.