ENHANCED METHODS FOR NATIONAL 3D GEOLOGICAL MODELLING IN CANADA

National 3D geological models and related knowledge-bases are cornerstones of geological frameworks underpinning a wide variety of applications in health, safety, sustainability, and the economy. Such framework development is a key objective for many government agencies as well as some academic efforts, however, it is impeded at national, continental, or global scales by methods and tools incapable of handling the volume and complexity of available information. To overcome this problem, the Canada3D project is developing new approaches to every stage of its workflow, in partnership with the international LOOP consortium (https://loop3d.org). Recent advances in AI, primarily in knowledge representation and machine-learning, are leveraged to enhance key workflow components:

• Knowledge extraction: in collaboration with associated projects, text mining approaches extract knowledge and data from published sources;
• Knowledge management: semantic approaches organize and store interpretative and textbook knowledge using a newly developed geoscience ontology;
• Data management: efficient methods store observations and measurements;
• 3D modelling: new machine-learning and agent-based approaches model large an complex geological terrains, and increasingly integrate with geophysical methods and tools;
• Model Validation: new knowledge-driven approaches check the consistency of a 3D model with known geological knowledge, to help filter unfeasible models, support iteration to a best fit, and eventually quantify uncertainty.
• Visualization: in collaboration with vendors, customized hierarchical approaches visualize massive subsurface volumetric representations at varying resolutions.

While all components are in various phases of development, their completion should lead to a dramatically new workflow for national 3D geological modelling in Canada, and result in significantly improved 3D models, particularly in large and complex geographic regions.