Surface and Borehole Ground Penetrating Radar for Advanced Prospection and Visualisation of Archaeological Targets

Abstract

The application of ground penetrating radar (GPR) methods to the imaging of archaeological targets has been widely documented over the last 20 years. In spite of many successful applications, the use of the GPR technique is still impeded when low signal-to-noise ratio (SNR) data are obtained. Such data quality may result when a) surveying over an electrically conductive subsurface, b) where a target has a low physical contrast with host media, or c) where subsurface reflectivity is structurally complex. Three-dimensional (3-D), multi-offset (MO) GPR surveying offers the potential to improve the GPR image of a target where such problems exist. The 3-D survey facilitates the process of 3-D migration, allowing complex structure to be correctly imaged. The MO method is a means of boosting signal amplitude, thereby improving signal-to-noise ratio (SNR) in low-signal data. The application of these techniques to the GPR setting is rare, but rarer still are documented examples of their integrated use. We therefore acquire archaeological GPR data using 3-D and MO surveying techniques to investigate the improvement to image quality when compared to a more conventional acquisition and processing strategy. GPR data were acquired in Spring 2006 over a known Romano-British target at Groundwell Ridge, near Swindon, UK, using a Sensors & Software pulseEKKO1000 system equipped with 450 MHz antennae. 3-D methods were facilitated by acquiring data with a sample density of 5×5 cm2. The MO acquisition was sampled with equal density over an area of the archaeological target where SNR was notably low and the interpretation of structure in archive datasets was cautious at best. The MO data were acquired to provide an expected factorial increase in SNR of 4·5 on processing. Use of 3-D methods significantly improves the spatial resolution of the output archaeological image; exterior walls are clearly imaged, together with the layout of interior rooms. The MO acquisition improves SNR such that the cautiously interpreted structure is more clearly imaged hence the interpretation can be made with more certainty. Further work is currently needed to quantify any image enhancement and to find a compromise between such detailed sampling and output image quality.