Spatial Interpolation Using Machine Learning: From Patterns and Regularities to Block Models

• DOI: 10.1007/s11053-023-10280-7
• Publication Date: 2023-11-25T14:01:34Z
• Authors: Glen T. Nwaila; Steven E. Zhang; Julie E. Bourdeau; Hartwig E. Frimmel; Yousef Ghorbani
• Keywords: Interpolation; Geostatistics

Abstract In geospatial data interpolation, as in mapping, mineral resource estimation, modeling and numerical geosciences, kriging has been a central technique since the advent of geostatistics. Here, we introduce new method for spatial interpolation 2D 3D using block discretization (i.e., microblocking) purely machine-learning algorithms workflow design. This paper addresses challenges patterns regularities nature, how different approaches have used to cope with these challenges. We specifically explore advantages drawbacks while highlighting long complex sequence procedures associated kriging. argue that techniques offer opportunities simplify streamline process mapping especially cases strong relationships between sample location concentration. To test method, synthetic were both geometallurgical porphyry Cu deposit. The very useful validating performance proposed microblocking able reproduce known values at unsampled locations. Our delivers benefits machine learning-based approach, which includes its simplicity (a minimum 2 hyperparameters), speed familiarity scientists. enables scientists working on employ workflows familiar their training, tackle problems previously solely domain geoscience. exchange, expect our will be gateway attract more scientist become geodata scientists, benefitting modern data-driven value chain.