Analysis and interpretation of GPR datasets for integrated archaeological mapping

An integrated approach to ground-penetrating radar interpretation should include not only the standard amplitude slice maps and isosurface renderings but also an analysis of individual reflection traces and adjusted and processed reflection profiles. Only when all those basic datasets are interpreted can the plethora of reflection features at various depths and locations within a grid be understood, especially in complex geological and archaeological settings. Topographically adjusted profiles can provide important clues to changes in reflectivity along a transect, indicating why certain amplitude features are visible (or not) in slice maps. An integration of excavation and outcrop data with reflection profiles can often indicate what features are producing high-amplitude reflections and which are yielding no reflection at all. Even individual reflection traces can be studied for polarity changes, which can help in identifying the types of buried materials that are producing reflections. All these datasets, some of which are often overlooked, must be integrated during interpretation, especially in complicated ground conditions. processing and analysis. This is a very positive development as the three-dimensional power of the GPR method is being appreciated and applied to many fields, including archaeology (Conyers 2012). Here I present a critique and reminder to those who use standard processing method for GPR, as many recent adherents have been ignoring the basic GPR reflection data in their interpretation. I propose that an integrated analysis is necessary using both the now-standard three-dimensional images with the basic data (reflection traces and profiles) from which these images are produced. When all GPR data are analyzed holistically, a better understanding of the often-complex three-dimensional output can be made. Examples are presented to show how standard software steps that slice datasets to produce threedimensional maps can sometimes produce misleading and erroneous interpretations. Topographic corrections prior to slice mapping can often alleviate some of these amplitude sampling problems, but in complexly layered ground, only a manual interpretation of reflection profiles will produce accurate conclusions. Sometimes, buried features of interest do not reflect radar energy at all, and therefore, a completely different interpretation that searches for low or absent reflection amplitudes is necessary. Often, only a detailed analysis of individual reflection traces can determine what types of materials in the ground are producing reflections of interest, with an example presented of reflected wave polarity showing differences in radar velocity in the ground as a function of physical and chemical properties along reflection interfaces. INTRODUCTION TO INTEGRATED GPR DATA PROCESSING The use of ground-penetrating radar (GPR) for archaeological mapping and interpretation has evolved from a purely exploratory technique using an interpretation of two-dimensional reflection profiles to one that now commonly uses three-dimensional mapping and computer-generated visualization programs to study much larger areas of the subsurface (Conyers 2013; Goodman and Piro 2013). These now-standard visualization techniques produce amplitude slice maps from two-dimensional reflection profiles and generate isosurface renderings from those complex three-dimensional datasets (Linford 2014) and recently from three-dimensional data collected using multi-antenna arrays that generate “real” three-dimensional output (Conyers and Leckebusch 2010; Novo et al. 2008; Trinks et al. 2010), including animations of large complex datasets. The use of multiple arrays also allows for three-dimensional migration of complex reflections, producing a much more “crisp” and accurate three-dimensional set of images (Sala and Linford 2012). All of these new collection and processing techniques are the product of robust and easily accessible hardware and software advances that can collect and process very large datasets quickly and efficiently. Many recent GPR adherents have joined the shallow geophysics community as GPR collection systems have become common, more intuitive to operate during data acquisition, and with efficient data transfer to powerful computers for