Probabilistic models for estimating lithology

The estimation of lithology within a volume of rock from surface mapping and borehole logging has been achieved using probabilistic models. The application of two geostatistical techniques (indicator kriging and conditional simulation) allows the incorporation of observed lithologies, taking cognisance of their statistical and spatial distributions. The newly developed probabilistic Bayes-Markov method additionally allows for the sequences of lithologies to be modelled by the calculation of transition probabilities and Markov chains. Verification of the estimates produced by these three methods has been enabled using the rocks actually exposed during the driving of an access tunnel into the ؤspِ Hard Rock Laboratory in south-east Sweden.