Rapid coal-ash characterization using geophysical methods & machine learning

Coal combustion products (CCP) are challenging to delineate in heterogeneous field settings. Conventional methods (test pits, coring, and laboratory analyses) are labor-intensive, slow, invasive, and provide sparse spatial coverage. This study evaluates whether rapid non-invasive geophysical screening methods—spectral induced polarization (SIP), magnetic susceptibility, and nuclear magnetic resonance (NMR) —combined with surface colorimetry ( RG B 24 ), can discriminate CCP-soil mixtures and provide reliable estimates of CCP content. Laboratory measurements were collected on five CCP-soil mixtures (series) and modeled using (i) a linear baseline, (ii) a calibrated non-linear (power-mean) model, and (iii) a machine-learning (ML) Random Forest approach, with validation via leave-one-series-out and site-specific tests. Across the five series, individual signals—particularly SIP and magnetic susceptibility—were strongly predictive of ash content but were consistently outperformed by combined models. The pooled calibrated non-linear and ML models captured the observed non-linearity and achieved high accuracy and precision, improving on linear fits. Colorimetry showed the weakest direct relationship with ash content but improved performance when included in multi-signal models. At a 3.5% decision threshold, calibrated and ML approaches yielded near-perfect classification (Matthews correlation coefficient ≈ 1), suggesting strong practical operability for field screening. Additionally, field-analog tests highlighted the role of endmembers—accuracy declined without access to endmember measurements but was largely recovered by collecting a minimal labeled pair for local recalibration. With end members, accuracy remained high. Globally trained models performed well on three operational unknowns; however, series-specific refits provided the most accurate predictions. Overall, these results highlight the potential of combining rapid geophysics and minimal local calibration for improved coal-ash delineation. Plain language summary Environmental remediation of legacy coal combustion product (CCP) impoundments can require excavation from unlined facilities. A persistent challenge is identifying boundaries of the CCP deposit and confirming when CCP has been fully removed. Field crews typically rely on knowledge of the CCP impoundment construction and visual inspection, supplemented in some cases by confirmation samples analyzed in a commercial lab with polarized light microscopy (PLM), or on-site colorimetric measurement. While testing of samples enables the use of independent analytical methods, sample testing inevitably results in sparse spatial coverage, and off-site analysis can result in delays in decision-making, and may necessitate costly remobilization if results reveal residual CCP. This study evaluates rapid screening signals from geoelectrical methods, magnetic methods, and color-value indices to determine whether ash-soil mixtures can be distinguished and ash content can be reliably estimated. The results demonstrate that each signal was informative on its own—however, combining signals, as well as integrating a machine-learning model, resulted in the best performance. This approach converts rapid measurements into reliable, decision-ready maps for excavation, which may ultimately help reduce costs, shorten timelines, and improve remedial outcomes. In the absence of endmember samples, the screening results still remain useful.