Using AI to Create Optical Stratigraphy - Higher Confidence Well Correlation and Determination of Reservoir Distribution

Observations of cuttings (i.e. colour, grain size and shape) are typically brief during drilling. These observations are useful but rarely comprehensive or consistent. This paper introduces an innovative approach using AI image segmentation. Based on a segment anything model it isolates and analyses thousands of individual grains in cuttings samples. The resulting data is stitched back together statistically, creating new forms of digital logs to support correlation, facies interpretation and data integration.

AI tools are emerging in rock image interpretation targeting specific problems within the mining sector. This tool built on the AI principle of “humans in the loop” ; keeping the geologist in the loop. Instead of providing an interpretation, the system delivers interpretable data (colour, grain size, aspect ratio, trace components, etc.) that can guide and support decision-making by domain experts.

One of the key areas of focus is the use of colour; a breakdown of grain-by-grain variations within an image, including bulk and trace contributions. Colour is mentioned in passing within geological descriptions and the human eye does not capture the full range and complexity of colour variations that can inform correlation decisions and determination of reservoir distribution.