Improved domaining for geostatistical modeling

地质统计学使用统计技术来评估地质矿床的资源/储量。这些技术在理解稀疏采样矿床、油藏和其他空间分布现象中的不确定性方面已经变得越来越流行。这项工作涉及更好地对表现出复杂地质特征的矿床进行建模的技术的开发；这些特征是使用定义特定岩石类型(如砂岩、页岩、花岗岩等)的岩石类型来建模的。通常，地质学家解释钻孔数据/图像以确定岩石类型，但数值模拟需要关于每种岩石类型的统计假设，根据地质定义，这些假设可能不正确。 计划研究的第一个方面是从统计学角度探讨岩石类型定义的建模含义。将根据现有的定量数据(如矿物品位、地球物理勘测、污染程度)和现有的定性数据(如地质解释)，为从矿藏中提取的每个样本分配一种岩石类型。这将确保岩石类型满足所有必要的建模假设，但也将保持从定性数据获得的地质知识的好处。 本研究的第二个方面是改进大比例尺矿体边界建模。这涉及到解释矿床的现有数据和确定矿化程度。通常情况下，这是利用矿床的地质知识来完成的；然而，提出了一种自动化方法作为地质解释的起点，以改进矿化模型。 这项研究一般针对所有需要空间建模的学科，包括但不限于：矿产资源/储量建模；采矿规划；污染建模；石油资源建模。然而，这项研究将在矿藏上进行演示。这项工作的预期结果是为样本数据分配类别(即岩石类型、相等)的方法、计算程序和建模建议，以及自动大规模矿化范围建模。 工程决策是根据这些数字模型作出的，包括：采矿计划；矿山的环境足迹；储存决定；工厂加工投入饲料。拟议工作的好处是考虑了已知的不确定性，并提高了数值模型的精度，从而改善了工程决策。这项研究给加拿大带来的好处是，由于对矿体进行了更好的建模，加拿大矿业公司将获得更大的竞争优势。将建立更准确的矿床模型，从而制定更好的采矿计划，增加加拿大采矿的利润和可持续性。

Geostatistics uses statistical techniques to assess resources/reserves for geological deposits. These techniques have become increasingly popular for understanding uncertainty in sparsely sampled mineral deposits, petroleum reservoirs and other spatially distributed phenomenon. This work relates to the development of techniques to better model deposits that exhibit complex geological features; these features are modeled using rock types that define a particular rock type (such as sand stone, shale, granite, etc). Usually a geologist interprets drill hole data/images to determine rock types but numerical modeling requires statistical assumptions about each rock type that may not be correct depending on the geological definition. The first aspect of the planned research is to explore the modeling implications of the definition of rock types from a statistical point of view. Each sample taken from an ore deposit will be assigned a rock type based on the quantitative data available (such as mineral grade, geophysical surveys, contaminate levels) and the qualitative data available (such as the geological interpretation). This will ensure that rock types meet all necessary modeling assumptions but will also maintain the benefit of geological knowledge from qualitative data. The second aspect of this research is to improve large scale ore body limits modeling. This involves interpreting the available data for a deposit and determining the mineralization extents. Normally this is done using the geological knowledge of the deposit; however, an automated method is proposed as a starting point for geological interpretation to improve models of mineralization. This research is generally directed towards all disciplines where spatial modeling is required, including but not limited to: mineral resource/reserve modeling; mine planning; contaminate modeling; petroleum resource modeling. However, the research will be demonstrated on mineral deposits. The anticipated outcomes of this work are methodologies, computational programs and modeling recommendations for the assignment of categories (i.e. rock types, facies, etc) to sample data as well as automatic large scale mineralization extents modeling. Engineering decisions are made based on these numerical models, including: mine plans; environmental footprints of mines; stockpiling decisions; plant processing input feeds. The benefits of the proposed work are to account for known uncertainties and increase the accuracy of numerical models, resulting in improved engineering decision making. The benefits of this research to Canada will be the increased competitive advantage for Canadian mining companies due to better modeling of ore bodies. More accurate models of ore deposits will be constructed, resulting in better mine plans with increased profits and sustainability of mining in Canada.