Quantitative Archaeological Feature Identification Using Handheld Spectrometers

Soil colour and texture play important roles in identifying archaeological features during excavations, particularly in rescue archaeology where rapid and reliable interpretation is required. This study investigated the application of visible-near-infrared (VIS-NIR) soil spectroscopy for quantitatively characterising cultural heritage materials and archaeological soils on freshly exposed surfaces after topsoil removal during excavation. Surface soil spectra were collected using a portable spectrometer from nine features at a rescue excavation site in Hyeondo-myeon, Republic of Korea. A PCA-based spectral deviation approach was applied to detect deviations of archaeological soils from locally defined natural background spectra. Balanced accuracy values exceeded 0.70 under optimised configurations across all sites, with several sites achieving values above 0.80. Strong statistical discrimination coincided with spatially coherent clustering of elevated anomaly values corresponding to archaeologically identified feature zones. The 400-1000 nm wavelength range, combined with locally calibrated background spectra, yielded the most stable and reproducible performance. The proposed workflow demonstrates that field-based VIS-NIR spectroscopy can provide rapid, quantitative, and spatially interpretable support for archaeological feature identification. By integrating sensor-based spectral characterisation with anomaly mapping, the approach minimises interpretive subjectivity and improves analytical reproducibility in excavation decision-making processes.