Assessing uncertainties in determining GPR ground wave velocities: An approach based on spectral velocity analysis

We propose a spectral velocity analysis approach to determine ground wave velocity from multi-offset or common midpoint (CMP) ground-penetrating radar (GPR) data. Our method is based on the linear moveout equation of ground wave arrivals and uses the unnormalized crosscorrelation to measure the coherency of the waveforms across the data gather. Peaks in the resulting velocity spectra indicate optimum velocities of linear events present in the data. Thus, our method is conceptually similar to CMP velocity analysis of reflected events where velocity spectra are computed using different coherency measures along hyperbolic trajectories. Furthermore, we propose a simple method to estimate uncertainties of the determined velocities. The introduced measure of uncertainty is based on the width of the maxima in the calculated velocity spectrum. Using different synthetic data examples, we test our approach and evaluate the influence of the true subsurface velocity, the surveying geometry, and the signal frequency on the accuracy of ground wave velocity estimates. Furthermore, we demonstrate the impact of these uncertainties on soil water content derived from these velocity values.