Random Field Models

This chapter contains definitions of classes of random fields suitable as probabilistic models for spatial continuous, event and mosaic variables. These respective classes are the Gaussian, Poisson and Markov classes. Prior models from these classes have conjugate properties with respect to likelihood models representing frequently used observation acquisition procedures for spatial variables. The model parameters may also be specified as random, which results in a hierarchical random field model. The conjugate characteristic is maintained if suitable prior pdfs for these parameters are assigned.