A new method for the identification of archaeological soils by their spectral signatures in the vis-NIR region

Abstract

This paper introduces a statistical method to identify spectral signatures of buried archaeological remains and distinguish them from spectra of the background soil in the visible to near infrared region. The proposed method is based on the Principal Component Analysis (PCA). The difference between an archaeological spectrum and non-archaeological soil spectra is quantified by a so-called R value. R values larger than 1 indicate that the spectrum represents an archaeological material. The method is successfully applied to samples from five study sites in Italy and Hungary with special conditions. The reflection spectra are taken in a time-efficient way with a field spectrometer. The method works best if background non-archaeological soil spectra are gathered from the same area, around the targeted archaeological site. Also, it can work without such local background spectra (but with lower accuracy) using background spectra from existing spectral libraries. This indicates that the method can, in principle, be applied to any archaeological site which is spectrally distinct from its surroundings. The paper highlights that this method does not require high spectral resolution and thus has the advantage of using low spectral resolution spectrometers which can eventually be applied for continuous 2D imaging applications with high temporal resolution. Additional studies are needed to further improve the method and to investigate under which conditions it works well or only with limited accuracy.