Datamix python package to process and integrate geospatial data

In geoscience many data are represented in the form of maps. Maps can be at various scales, the size of study domain, region, country or global. Maps are also often the result of quantitative or qualitative research or evaluation study. Maps are very effective in geoscience to synthetize or aggregate various results or data spatially.

Map construction involves creating, visualizing, and analyzing maps through interpolation, extrapolation, and regression. Highly multivariate and non-linear statistical correlations can be captured with Machine Learning techniques, and combined with more classical geostatistical methods to include spatial autocorrelation, geological knowledge and further calibration to well data. A comprehensive workflow is essential for managing the multi-step process of producing maps efficiently.

Datamix is a Python library designed for geoscience map operations, offering tools for building, displaying, and analyzing data maps, as well as performing geometric calculations and multi-criteria decision analysis such as AHP (Analytic Hierarchy Process). It also features a workflow system for automating map-building steps, enabling probabilistic predictions from stochastic analyses.

The Datamix package, developed in the Tellus project, has been applied to evaluate the geothermal potential of British Columbia using the AHP method. This involved creating a prospectivity map by combining criteria maps, such as heat flow, faults, volcanoes, and hot springs, with expert assessments based on geological domains. The package was also used to assess CO2 storage potential in the Balder Formation of the Norwegian North Sea. This study involved constructing a porosity map using machine learning and applying the AHP method to evaluate criteria such as porosity, burial, thickness, fault distance, and net-to-gross ratios.

The workflow demonstrated in the two use cases is adaptable for creating various types of prospectivity maps with different datasets. The Datamix package is versatile for other geoscience applications and can be expanded with newly developed methods to enhance its functionality.