Adoko, Collins Gogoe2024-06-272024-06-272024-04-17Adoko, C.G. (2024). An Innovative Truncated Gaussian Collocated Co-simulation Approach for Stochastic Modeling of Geological Domains. Nazarbayev University School of Mining and Geoscienceshttp://nur.nu.edu.kz/handle/123456789/8041In order to get a precise assessment of the quantity of mineral resources contained inside a deposit, it is necessary to create a comprehensive model that encompasses its geological domains. Efficient modeling of these geological domains (rock units, alteration types, and mineralization zones) is vital as they often serve as the mineralization controls of the deposit. Deterministic modeling tools such as wireframing make geological assumptions, especially in the intervals between drill holes, and so may yield incorrect conclusions about the complexity of the deposit in such intervals. Moreover, the uncertainty of rock units in unsampled places cannot be quantified using these deterministic modeling techniques. As such, methods for stochastic modeling are typically favored. The two most often used stochastic approaches that are better suited to address these shortcomings of the deterministic modeling of geo-domains are sequential indicator simulations and truncated/plurigaussian Gaussian simulations. The Truncated Gaussian simulation is an essential simulation approach for defining complicated geometry of rock units because it can accurately represent the spatial connection between rock units and quantify their uncertainty. However, long-scale geological structures like veins, faults, and fractures tend to be loosely modeled when employing this approach; in order to prevent this inaccuracy, soft data such as deterministic interpretive geological models can be used. In this project, the variability of rock units in the subsurface is assessed and the uncertainty in their occurrences in a gold deposit which is vein-dominated is quantified using the Truncated Gaussian simulation. A new method for modeling the rock units—in particular, the long-scale geological structures (veins in this case) is provided. The method involves integrating the conventional Truncated Gaussian simulation with a collocated co-simulation algorithm that is based on the inclusion of a local and global correlation coefficient parameter. The realizations are conditioned to data from drill holes and a machine learning algorithm-generated collated interpretive geological model. A number of realizations, 100 in total, are produced from both the local and global correlation co-simulations. The results show the practical spatial representation of the veins and other rock units as validated by the information from drill holes and also appropriately quantify the uncertainty associated with their occurrences compared to the conventional Truncated Gaussian.enAttribution-NonCommercial-NoDerivs 3.0 United StatesType of access: RestrictedMaster's thesis