Geostatistical applications in petroleum geology and sedimentary geology

Statistical tools can aid the geologist in understanding geologic data, and in the solution of real-world problems. This thesis examines several tools, concentrating on techniques with wide applicability in sedimentary and petroleum geology.; A case study presents the use of geostatistics for mapping hydrocarbon pore volume and risk assessment in an area surrounding Amos Draw field in the northern Powder River basin of Wyoming. The study used sequential Gaussian and indicator simulation techniques, and documents the ability of indicator simulation to incorporate correlated secondary data.; Reservoir characterization requires the generation of numerical grids of geologic properties. Because those properties differ for each rock type, one should first simulate the distribution of rock types, and then the distribution of the reservoir properties. This thesis proposes two multivariate statistical techniques, discriminant function analysis and cluster analysis, for the identification of petrophysical rock types in the Muddy Formation at Amos Draw. The geology of those rock types is discussed using core descriptions, thin-sections, and well log data. The rock types were simulated in three dimensions using indicator principal component simulation. The study also used simulated annealing for post-processing of the simulations, incorporating information from the wells on the transition frequencies between the rock types.; The third case study used runs analysis for the identification of patterns in bed thickness and grain size in turbidites. Upward-thickening and -thinning patterns have been used to assign turbidite sequences to depositional environments, although there has been disagreement on their identification. Runs analysis was applied to a turbidite section in the Sites Formation at Cache Creek, in northern California. That sequence has been described as a classic example of upward-thickening cycles in a progradational outer fan lobe. Runs length distributions show that the sequence consists of subequal numbers of upward-thinning and -thickening cycles within an overall upward-thickening and -fining sequence. Therefore, the sequence does not appear to be the result of fan lobe progradation. By quantifying and comparing the patterns of sedimentary features observed in turbidite sequences, it may be possible to generate new depositional models in the future that reflect the true variability in deep-sea clastic environments.