Item

Abstract

Colour of buried archaeological remains tends to be different from the adjacent soils due to anthropogenic activities influencing the chemical properties of the soil. Such distinctive colour difference indicates that buried remain might be easily identified by their colour, but the colour recognised by the human eye (characterised by the contributions of light in the red, green and blue spectral regions) is not always sufficient to clearly distinguish between natural soils and archaeological artefacts. This thesis attempts to use the full information content of reflectance spectra in the visible and near infrared spectral range to identify archaeological remains. Since such reflectance spectra are attenuated in a complex way by scattering and absorption processes, the spectral characteristics of archaeological remains are investigated using a modified principal component analysis (PCA) method. The PCA method is extended in a way which allows to quantify the differences between a spectrum of interest and a group of selected natural soils by a distance value ‘D’. Large D values indicate that the spectrum represents a non-natural soil, e.g. an archaeological material. Archaeological sites investigated in this thesis have provided positive results (large D values) for pits, ditches and archaeological features influenced by fire activity. The developed method works best if reference spectra of local soils are used, but even with a global database of soil spectra, archaeological materials can still be identified. These results indicate that the method can be applied in a universal way to any archaeological sites. However, this hypothesis has to be further proven by additional investigations. It should also be noted that the developed method might not only be used for archaeological applications but also to distinguish in a general sense whether the soil colour difference is due to natural process or anthropogenic influence. Another important result of this study is that spectral features of archaeological remains can still be identified with much lower spectral resolution than provided by the spectrometer used (3 – 10 nm). These findings demonstrate a promising approach to use instruments with coarser spectral resolution, but largely improved temporal resolution, which might even allow continuous 2D imaging applications.