Theoretical generalization of Markov chain random field from potential function perspective

Abstract

The inner relationship between Markov random field (MRF) and Markov chain random field (MCRF) is discussed. MCRF is a special MRF for dealing with high-order interactions of sparse data. It consists of a single spatial Markov chain (SMC) that can move in the whole space. Generally, the theoretical backbone of MCRF is conditional independence assumption, which is a way around the problem of knowing joint probabilities of multi-points. This so-called Naive Bayes assumption should not be taken lightly and should be checked whenever possible because it is mathematically difficult to prove. Rather than trap in this independence proving, an appropriate potential function in MRF theory is chosen instead. The MCRF formulas are well deduced and the joint probability of MRF is presented by localization approach, so that the complicated parameter estimation algorithm and iteration process can be avoided. The MCRF model is then applied to the lithofacies identification of a region and compared with triplex Markov chain (TMC) simulation. Analyses show that the MCRF model will not cause underestimation problem and can better reflect the geological sedimentation process.

Keywords

localization approach / Markov model / potential function / reservoir simulation / transiogram fitting

References

[1] VisteliusA B. On the question of the mechanism of formation of strata [J]. Doklady Akademii Nauk SSSR, 1949, 65(2): 191-194

[2] KrumbeinW C, DaceyM F. Markov chains and embedded Markov chains in geology [J]. Mathematical Geology, 1969, 1(1): 79-96

[3] CarleS F, FoggG E. Modeling spatial variability with one and multi-dimensional continuous-lag Markov chains [J]. Mathematical Geology, 1997, 29(7): 891-918

[4] KrumbeinW CFORTRAN IV computer program for simulation of transgression and regression with continuous-time Markov models [D], 1968Evanston, IllinoisNorthwestern University

[5] LinC, HarbaughJ WGraphic display of two and three dimensional Markov computer models in geology [M], 1984New YorkJohn Wiley & Sons, Inc180

[6] ElfekiA, DekkingM A. A Markov chain model for subsurface characterization: Theory and applications [J]. Mathematical Geology, 2001, 33(5): 569-589

[7] GrayA J, KayJ W, TitteringtonD M. An empirical study of the simulation of various models used for images [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1994, 16(5): 507-513

[8] SalomãOM C, RemacreA Z. The use of discrete Markov random fields in reservoir characterization [J]. Journal of Petroleum Science and Engineering, 2001, 32(2): 257-264

[9] NorbergT, RoséNL, Baran, BaranS. On modelling discrete geological structures as Markov random fields [J]. Mathematical Geology, 2002, 34(1): 63-77

[10] LiuZ-f, HaoT-y, WangFeng. Stochastic simulating distribution of continuous reservoir attributes based on the GMRF model [J]. Petroleum Exploration and Development, 2005, 32(6): 75-81

[11] LiangY, WangZ, GuoJ. Reservoir lithology stochastic simulation based on Markov random fields [J]. Journal of Central South University, 2014, 21(9): 3610-3616

[12] LiW. Markov chain random fields for estimation of categorical variables [J]. Mathematical Geology, 2007, 39(3): 321-335

[13] LiW. A fixed-path Markov chain algorithm for conditional simulation of discrete spatial variables [J]. Mathematical Geology, 2007, 39(2): 159-176

[14] LiW, ZhangC. Some further clarification on Markov chain random fields and transiograms [J]. International Journal of Geographical Information Science, 2013, 27(3): 423-430

[15] BillingsleyPProbability and measure [M], 19953New YorkJohn Wiley & Sons, Inc515

[16] LiW, ZhangC, BurtJ E, ZhuA, FeyanJ. Two-dimensional Markov chain simulation of soil type spatial distribution [J]. Soil Science Society of America Journal, 2004, 68(5): 1479-1490

[17] BesagJ. Spatial interaction and the statistical analysis of lattice systems [J]. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 1974, 36(2): 192-236

[18] LiS ZShanbhagD N, RaoC R. Modeling image analysis problems using Markov random fields [M]//. Handbook of Statistics 21: Stochastic Processes: Modeling and Simulation, 2003473-513

[19] HAMMERSLEY J M, CLIFFORD P. Markov fields on finite graphs and lattices [EB/OL]. [1971]. http://www.statslab.cam.ac.uk/~grg/books/hammfest/hamm-cliff.pdf.

[20] GriffeathDKemenyJ G, SnellJ L, KnappA W. Introduction to random fields [M]//. Denumerable Markov Chains, 1976New YorkSpringer425-458

[21] StienM, KolbjØRnsenO. Facies modeling using a Markov mesh model specification [J]. Mathematical Geosciences, 2011, 43(6): 611-624

[22] RipleyB DGriffithD A. Gibbsian interaction models [M]//. Spatial statistics: Past, present and future, 1990New YorkInstitute of Mathematical Geography.1-19

[23] LiW, ZhangC. Simulating the spatial distribution of clay layer occurrence depth in alluvial soils with a Markov chain geostatistical approach [J]. Environmetrics, 2010, 21(1): 21-32

[24] ZhangC, LiW. Comparing a fixed-path Markov chain geostatistical algorithm with sequential indicator simulation in categorical variable simulation from regular samples [J]. GIScience & Remote Sensing, 2007, 44(3): 251-266

[25] LiW. Transiogram: A spatial relationship measure for categorical data [J]. International Journal of Geographical Information Science, 2006, 20(6): 693-699

[26] LiW, ZhangC. A generalized Markov chain approach for conditional simulation of categorical variables from grid samples [J]. Transactions in GIS, 2006, 10(4): 651-669

[27] LiW, ZhangC. Application of transiograms to Markov chain simulation and spatial uncertainty assessment of land-cover classes [J]. GIScience & Remote Sensing, 2005, 42(4): 297-319

[28] ZhangC, LiW. Markov chain modeling of multinomial land-cover classes [J]. GIScience & Remote Sensing, 2005, 42(1): 1-18

[29] LeederM RSedimentology: process and product [M], 1982LondonGeorge Allen & Unwin Ltd344

[30] DovetonJ HYarusJ M, ChambersR L. Theory and applications of vertical variability measures from Markov chain analysis [C]//. Stochastic modeling and geostatistics: Computer applications in geology, 1994

[31] LiJ, YangX-j, ZhangX-l, XiongL-ping. Lithologic stochastic simulation based on the three-dimensional Markov chain model [J]. Acta Petrolei Sinica, 2012, 33(5): 846-853

[32] MaY Z, JonesT A. Teacher’s aide: Modeling hole-effect variograms of lithology-indicator variables [J]. Mathematical Geology, 2001, 33(5): 631-648

[33] DeutschC V, JournelA GGSLIB: Geostatistics software library and user’s guide [M], 19922New YorkOxford University Press340

[34] JournelA G, FroidevauxR. Anisotropic hole-effect modeling [J]. Mathematical Geology, 1982, 14(3): 217-239

[35] HohnM EGeostatistics and petroleum geology [M], 1988New YorkVan Nostrand Reinhold Company264

[36] ChristakosGRandom field models in earth sciences [M], 19922San Diego, CAAcademic Press474

[37] LiW, ZhangC, DeyD K. Modeling experimental crosstransiograms of neighboring landscape categories with the gamma distribution [J]. International Journal of Geographical Information Science, 2012, 26(4): 599-620

[38] CarleS F, FoggG E. Transition probability-based indicator geostatistics [J]. Mathematical Geology, 1996, 28(4): 453-476