Advanced Geophysical Processing of Ground-Penetrating Radar and Magnetic Datasets

Abstract

Due to the presence of an excessive amount of noise in the data, archeological geophysics frequently produces results that cannot be used to evaluate the content that may exist in the subsurface. If it is impossible to differentiate between signal and noise, excessive noise will result. Its genesis, when it is attributable to heterogeneities in the ground (overthrows, corners) that produce as many reflections as structures that may exist, prevents a fair assessment of the subsoil composition. Low perceptibility circumstances happen when there is no contrast between buried structures and the surrounding environment. It could be due to elements that are harmful to the method (metals and ceramics in the magnetic method; clay and water in the electromagnetic method) or when the buried objects are formed of the same material as the surrounding medium. The problem of the identification and selection of useful signals in geophysical data is one of the research topics of effective methodologies of archeological geophysics, with the aim of producing more accurate models that allow a more precise interpretation of structures buried beneath the earth. In this paper, three approaches conceived by the team are described, which allow: to increase the sharpness of the GPR models by reducing the background noise through factoring techniques applied in the 2-D spectral domain; to increase the resolution of the models by increasing the density of the profiles with Fourier interpolation; and to increase model information by combining maps from the two geophysical methods, using data fusion techniques that combine mathematical transformations and statistical analysis.