Geoscientific modelling has some unique requirements. Modern geological applications require increasingly quantitative and accurate rock property characerizations within the three-dimensional subsurface environment. this problem differs from that faced by most other fields due to a variety of technocal and economic constriants. Three-dimensional geoscientific modelling often relies on complex stochastic concepts and thus requires the extraction of information from large multiparameter data sets, and the representation and modification of complex, and uncertain geo-objects of interest.