pyCoreRelator: A Quantitative Tool for Automated Core and Well-Log Correlation

Join Zoom Meeting
Meeting ID: 932 0137 7630
Passcode: 013011

In Person
Building 196/ROC 1.603

Presenter

Dr. Larry Syu-Heng Lai
Postdoctoral Researcher
Bureau of Economic Geology
Jackson School of Geosciences
The University of Texas at Austin

Description

Physical stratigraphic correlation between cores or well-logs is fundamental to basin analysis and sedimentological studies. However, objectively assessing the confidence of these correlations remains challenging, particularly across significant distances or different depositional environments. To address this, I developed a new Python tool, pyCoreRelator, designed to systematically identify all feasible unit-to-unit correlation solutions between data pairs while honoring available datums and accounting for lateral bed thickness variability. The tool employs the dynamic time warping (DTW) algorithm to align log data and compute similarity metrics. To determine if correlation solutions are geologically meaningful or coincidental, the tool assesses whether these metrics differ statistically from those of randomly stacked synthetic stratigraphy, enabling a robust evaluation of correlation quality. To test the effectiveness of this new approach, I applied pyCoreRelator to Cascadia offshore sediment cores to study the correlatability of marine turbidites previously interpreted as synchronous, earthquake-triggered deposits, using physical property logs and image color profiles in our analysis. For each core pair, the algorithm seeks optimal solutions by minimizing lateral changes in unit thickness (e.g., pinch-outs) and maximizing compatibility with age constraints. My findings show that correlation quality improves markedly for core pairs collected in proximity and within the same depositional environment. Notably, the correlation strength of turbidites over long distances and across different depositional settings is not improved, and is sometimes degraded, by including age constraints, calling into doubt the long-distance correlatability of some Cascadia turbidites and their paleoseismology implications. These results demonstrate that pyCoreRelator provides a robust method to find optimal correlation solutions compatible with geological constraints and to enable a reproducible assessment of correlation confidence. While further testing is necessary in other settings (e.g., carbonates, mixed systems), this work underscores the wide applicability of this approach for enhancing the objectivity, efficiency, and productivity of stratigraphic analysis.