Common‐ and multi‐offset ground‐penetrating radar study of a Roman villa, Tourega, Portugal

A ground-penetrating radar (GPR) survey was conducted at the site of the first to fourth century AD Roman villa of Tourega, near Evora, Alentejo region, Portugal. This site has been partially excavated, revealing a bathhouse complex consisting of a tank reservoir, multiple bathing rooms and a major corridor. The excavated portion of Tourega provides direct archaeological feedback as to the size, shape, depth and orientation of structures revealed by GPR surveys. A 4500 m2 area surrounding the known site was surveyed using common-offset 500 MHz and 200 MHz pseudo-three-dimensional acquisition. Amplitude time-slice analysis reveals a large number of well-defined additional structures confirming a much broader extension of the site. For shallow buried rectilinear structural targets, the high frequency 500 MHz common-offset data provide excellent imaging. However, limited depth penetration at 500 MHz necessitated a lower frequency of investigation; therefore 200 MHz common- offset and multi-offset data were also acquired. Two 200 MHz multi-offset GPR profiles were obtained over the large double-walled structure identified on the time-slices, acquired using a newly proposed multi-offset acquisition technique – rapid multi-offset acquisition. This acquisition technique uses the existing single-channel GPR system, is extremely cost-effective, and easily acquires a high fold. Rapid multi-offset acquisition provides an average time saving up to five times that of stationary multi-offset profiling. The 200 MHz multi-offset profiles provide a greater depth penetration and enhanced structural detail (i.e. improved imaging of dipping structures) than comparable 200 MHz common-offset profiles. Post-stack, inverse-Q filtering increases vertical resolution and interpretability of the multi-offset sections via removal of wavelet dispersion. Although multi-offset data are less feasible for characterizing this site in three dimensions because of the large time commitment in data collection, two multi-offset profiles over key structures of interest improved the final site interpretation, with enhanced image quality and higher resolution than comparable common-offset profiles. The combined results of the multi-offset survey with a common-offset grid should encourage a more frequent use of multi-offset data at archaeological sites when the depth extent of the archaeological site is uncertain. Copyright © 2007 John Wiley & Sons, Ltd.