Hotelling's T² confirms REE distribution supporting sustainable land application

Coal ash from stratigraphically complex basins shows compositional variability that can complicate safe land application, yet seam-level geochemical characterization remains limited. This study evaluated whether rare earth element (REE) fingerprints can reliably distinguish coal seams in the Pulau Balang Formation, Kutai Basin, East Kalimantan, Indonesia, and whether the resulting classification supports a tiered screening framework for mine-land rehabilitation. Forty-five coal ash samples from three stratigraphic seam groups (Upper: Seams 1-4; Middle: Seams 9-12; Lower: Seams 17-19 were analyzed for light REEs (La, Ce, Pr, Nd, Sm) by ICP-OES and for major oxides by XRF. REE data were log10-transformed and z-standardized before principal component analysis (PCA), linear discriminant analysis (LDA) with leave-one-out cross-validation (LOOCV), and pairwise Hotelling’s T² testing. Major oxides were treated as compositional data using the centered log-ratio (clr) transformation. PCA explained 79.3% of REE variance on PC1, with La loading separately from the Ce-Pr-Nd-Sm cluster. LDA achieved 100% LOOCV classification accuracy, and Hotelling’s T² confirmed significant multivariate separation across all seam pairs (p<1 × 10-15). In contrast, clr-transformed oxides separated only the Middle and Lower seams (p = 0.0010). Generally, REE fingerprinting provides a stronger basis for seam-source assignment than major oxide profiling and supports seam-based screening for selective stockpile management and controlled coal-ash placement in reclaimed mine lands.