The potential of hyperspectral and multi-spectral imagery to enhance archaeological cropmark detection: a comparative study

摘要

Aerial photography has made the single most important contribution to our improved appreciation of the density, diversity and distribution of archaeological sites in Britain since World War Two. This is particularly the case for areas of intensive lowland agriculture where ploughed-out sites are known mainly from marks in the crops growing above them. However, reconnaissance for such cropmarks is not equally effective throughout the lowlands, because of the particular conditions of drier weather, well-drained soils and arable agriculture required before they become visible, and is highly unpredictable. Given that the appearance of cropmarks is linked to moisture stress in growing plants, they are potentially detectable at bandwidths outside the visible spectrum and before they become apparent therein. This paper focuses on the application of two spectral enhancement techniques: Principal component analysis and Tasselled cap transformation. Comparing a range of imagery (CASI-2, ATM and digital vertical photographic data) from two case study areas in Lowland Scotland, each with very different environmental, agricultural and archaeological backgrounds to facilitate further comparisons, the paper demonstrates that multi-spectral/hyperspectral imagery can enhance the identification of otherwise invisible archaeological sites, particularly in the near-infrared part of the spectrum. However, the lower spatial resolution of such imagery, compared to photography, can make the often diffuse and incomplete cropmark traces more difficult to determine with confidence.