Archived
This content is available here strictly for research, reference, and/or recordkeeping and as such it may not be fully accessible. If you work or study at University of Kentucky and would like to request an accessible version, please use the SensusAccess Document Converter.
Location
Lexington, Kentucky
Start Date
6-5-2026 8:30 AM
End Date
6-5-2026 9:00 AM
Description
Groundwater monitoring networks at coal combustion residual (CCR) facilities now yield dataset volumes that routinely exceed the capacity of traditional, univariate interpretation. This presentation introduces a scalable multivariate and machine learning (ML) framework that prioritizes understanding data structure before predictive models are applied. The workflow applies Principal Component Analysis and other multivariate statistics to groundwater datasets, to illustrate how unsupervised learning can rapidly reveal dominant sources and controls on impacted and unimpacted groundwater quality. The talk frames multivariate analysis as a practical portfolio management tool, emphasizing inference categories rather than concentrations or site identifiers. Fleet-scale multivariate screening offers strategic advantages for CCR programs, including: • Clearer differentiation between natural background and CCR-influenced groundwater signatures, • Improved ability to assess the effectiveness of corrective action, • Earlier recognition of sites experiencing transitional or mixed geochemical behavior; and, • Improved cross-team analytical consistency as workflows are applied across multiple facilities. This workflow directly supports conceptual site model refinement and update, target well prioritization, and builds an analytical foundation for ML including exceedance risk forecasting and remedy-performance trend detection. The approach integrates directly into CCR groundwater management planning by enabling consistent monitoring logic and reducing interpretive uncertainty. The outcome is a scalable path from groundwater data structure to data-driven CCR compliance strategy, corrective action planning, and ML-readiness diagnostics.
Document Type
Presentation
Archival?
Archival
Included in
Energy Systems Commons, Environmental Indicators and Impact Assessment Commons, Environmental Monitoring Commons, Mining Engineering Commons, Oil, Gas, and Energy Commons, Structural Materials Commons, Sustainability Commons
Innovative Analysis of Groundwater Data for Scalable, Defensible, and Predictive Compliance
Lexington, Kentucky
Groundwater monitoring networks at coal combustion residual (CCR) facilities now yield dataset volumes that routinely exceed the capacity of traditional, univariate interpretation. This presentation introduces a scalable multivariate and machine learning (ML) framework that prioritizes understanding data structure before predictive models are applied. The workflow applies Principal Component Analysis and other multivariate statistics to groundwater datasets, to illustrate how unsupervised learning can rapidly reveal dominant sources and controls on impacted and unimpacted groundwater quality. The talk frames multivariate analysis as a practical portfolio management tool, emphasizing inference categories rather than concentrations or site identifiers. Fleet-scale multivariate screening offers strategic advantages for CCR programs, including: • Clearer differentiation between natural background and CCR-influenced groundwater signatures, • Improved ability to assess the effectiveness of corrective action, • Earlier recognition of sites experiencing transitional or mixed geochemical behavior; and, • Improved cross-team analytical consistency as workflows are applied across multiple facilities. This workflow directly supports conceptual site model refinement and update, target well prioritization, and builds an analytical foundation for ML including exceedance risk forecasting and remedy-performance trend detection. The approach integrates directly into CCR groundwater management planning by enabling consistent monitoring logic and reducing interpretive uncertainty. The outcome is a scalable path from groundwater data structure to data-driven CCR compliance strategy, corrective action planning, and ML-readiness diagnostics.
