GPR spectral analysis for clay content evaluation by the frequency shift method

• The GPR peak frequency moves towards lower frequencies, as clay content increases.
• GPR spectra are affected by low resolution.
• We apply super-resolution methods in the GPR frequency domain.
• Sub-sample resolution techniques refine the estimate of the frequency peak.
• The peak location is found through sinc-interpolation in the frequency domain.

A recent approach relates the shift of the frequency peak of the Ground Penetrating Radar (GPR) spectrum with the increasing of the moisture content in the soil. The weakness characterizing this approach is represented by the needs of high resolution signals, whereas GPR spectra are affected by low resolution. The novelty introduced by this work is twofold. First, we evidence that clay content information is present in the location where the maximum amplitude of the GPR spectra occurs. Then, we propose three super resolution methods, namely parabolic, triangular, and sinc-based interpolators, to further refine the location of the frequency peak. In fact, it is really important to be able to find this location quite precisely, to obtain accurate estimates of clay content. We show that the peak location can be found best through sinc-interpolation in the frequency domain of the measured data. Our experimental results confirm the effectiveness of the proposed approach to resolve a frequency shift in the GPR spectrum, even for a small amount of clay.