Airborne Spectral Imagery for Archaeological Prospection in Grassland Environments-An Evaluation of Performance
Bennett, Rebecca, Welham, Kate, Hill, Ross A., Ford, Andrew, Antiquity
The use of multispectral data in the form of satellite imagery is a relatively well-established technique for archaeological prospection and has been shown to be a powerful tool for site recognition and landscape scale analysis (Philip et al. 2002; Beck et al. 2007; Turner & Crow 2010). Applications of satellite data vary from palaeogeography (Khadkikar et al. 2004) to the identification of settlement areas in support of field investigations (Gheyle et al. 2004; Beck et al. 2007) or the looting of sites following the Iraq war (Stone 2008). However, to date, the use of airborne spectral sensors in historic environment research has been very limited, which is perhaps surprising given both the finer spectral and spatial resolution they can provide when compared with satellite-based sensors and their widespread use in the environmental sciences to detect vegetation and soil properties (Govender et al. 2007; Xie et al. 2008; Ben-Dor et al. 2009; Mulder et al. 2011).
Airborne spectral data was first used to detect archaeological remains over 20 years ago (Donoghue & Shennan 1988), since when a number of projects have shown the potential for the use of airborne digital spectral imagery for detection of archaeological features and its complementarity to other airborne and ground-based survey techniques (Winterbottom & Dawson 2005; Challis & Howard 2006; Powlesland et al. 2006; Challis et al. 2009). An emerging body of archaeological research into identifying features through soil spectral response is being undertaken in the Mediterranean region and the Fertile Crescent (Ben-Dor et al. 2002; Traviglia 2005; Rowlands & Sarris 2007). However, despite showing some promise, digital spectral imaging has not been widely applied by historic environment professionals, and research in the UK to date has focused almost exclusively on arable landscapes. For example, while digital spectral data was collected for the Stonehenge World Heritage Site, its analysis was secondary to that of airborne laser scanning (ALS or lidar survey) (Bewley et al. 2005). The single published example of the analysis of airborne , spectral data for the non-arable, coastal Machair environment of Coll and Tiree, Argyle and Bute, Scotland (Winterbottom & Dawson 2005) showed the potential of these data for identification of archaeological features but was not able to provide a quantitative comparison with the results of aerial photographic transcription. Therefore while the identification in airborne spectral data of crop marks in cereals has been illustrated, there has been little assessment of the performance of this type of sensor in comparison to established techniques. This is particularly true for non-arable environments where the change in vegetation properties caused by surface or sub-surface features can be more subtle than in landscapes dominated by arable cultivation.
This paper assesses the potential use of archive digital spectral data for the detection of archaeological features in non-arable landscapes using a case study on the calcareous grassland of Salisbury Plain, Wiltshire, UK. The research makes use of archive spectral data collected in 2002 by the Environment Agency of England and Wales (EA) on behalf of the Ministry of Defence for the purposes of land-use mapping (Barnes 2003). The results are compared with those of the National Mapping Programme's (NMP) transcription of archive aerial photography.
Airborne digital spectral imaging
The identification of changes in vegetation patterns as proxy indicators for archaeological features is a long-held tenet of aerial photographic survey and transcription. Vegetation (crop) marks, along with soil marks and topographical anomalies, can be mapped using standard monochrome or colour photography, given favourable conditions. However, standard photographic techniques are only sensitive to a small portion of the visible electromagnetic spectrum of reflected energy at wavelengths between 450-650nm. …