Evaluation of GPR Data Attributes to Investigate a Mud-Wall Building at a First Millennium ad Village, Catamarca, Argentina

Abstract

Detecting and mapping mud structures by using ground penetrating radar (GPR) is often challenging because of low target to background contrast and irregular surfaces of the targets. These characteristics tend to produce unclear signals of the mud structures in the conventional images of the data. In these cases, calculating attributes of the data could be useful to produce alternative images that facilitate or improve the interpretation. In this article, we evaluate the capabilities and deficiencies of a broad set of attributes of the GPR data for studying a mud-wall building, which was detected in a dry environment in the Andean region of north-western Argentina by using this method. This building, later interpreted as a dwelling, was inhabited during the 1st and 2nd centuries ad, dating back to the formative period of the region. The analysed attributes belong to four types, which are probably the most used in GPR: instantaneous, geometric, texture and interval attributes. Image characteristics that determine the visibility of the wall signals, such as contrast between the mean values of the property at the wall positions and around them, signal continuity along the walls and signal sharpness are discussed. The visibility of the wall signals is analysed as a function of the input parameters and the time coordinate. The main characteristics of the images are interpreted in terms of soil sources that probably produce them. Different groups of attributes are determined based on their individual abilities to improve the images. The capabilities, limitations and usefulness of calculating these attributes to detect the mud walls of the building and generate a plan view of them are shown. Including attributes in the analysis of the data allowed identifying the mud walls easier and more confidently than not doing so, as well as determining their layout more completely.