Markov chain geostatistics

From Wikipedia, the free encyclopedia

Markov chain geostatistics applies Markov chains in geostatistics for conditional simulation on sparse observed data; see Li et al. (Soil Sci. Soc. Am. J., 2004), Zhang and Li (GIScience and Remote Sensing, 2005) and Elfeki and Dekking (Mathematical Geology, 2001). There are also other two ways using transition probabilities (not truly Markov chains) for conditional simulation. One is the transition probability-based geostatistics, suggested by Carle and Fogg (Mathematical Geology, 1996). This approach is essentially an extension of indicator kriging, because they replace indicator variograms by transition probabilities in indicator kriging to generate simulated realizations. Another way is using Markov random field with simulated annealing, see T. Norberg et al. (Mathematical Geology, 2002).