Field observations are tested against modal propagation theory to find the practical limitations upon derivation of layer permittivities and signal attenuation rates from a radar moveout profile over two-layer ground. A 65-MHz GPR pulse was transmitted into a 30-60-cm-thick surface waveguide of wet, organic silty to gravelly soil overlying a drier refracting layer of sand and gravel. Reflection profiles, trench stratigraphy, resistivity measurements, and sediment analysis were used to quantify the propagation medium and possible attenuation mechanisms.
Highly dispersive modal propagation occurred within the waveguide through 35 m of observation. The fastest phase velocity occurred at the waveguide cutoff frequency of 30 MHz, which was well received by 100-MHz antennas. This speed provides the refractive index of the lower layer, so the near-cutoff frequencies must match a lower layer refraction. A slower, lower frequency phase of the dispersed pulse occurred at about 60–70 MHz, with an average attenuation rate of about 0.4 dB/m. Similar events appear to have reflected back and forth along the waveguide. Modal theory for the average layer thickness shows all primary events to be different aspects of a TE1 mode, predicts the correct 30–70-MHz phase speeds and low-frequency cutoff phenomenon, but also predicts that the 60–70-MHz group speed should be slightly lower than observed. An Airy phase was apparently out of the bandwidth. Two-dimensional finite-difference time-domain modeling qualitatively simulates the main field results.
After accounting for an inverse dependency of amplitude on the square of the range, the high resistivity of the surface layer accounts for the 0.4-dB/m attenuation rate for the 60–70-MHz phase of the pulse. However, erratic amplitudes, interface roughness, and the reflected packets indicate scattering. We conclude that permittivities can be well estimated from dispersive moveout profiles given an average surface layer thickness, and the wide bandwidth of GPR antennas allows the full dispersion to be seen. Attenuation rates appear to be derivable from the higher frequency part of our dispersive event, for which attenuation might be least affected by the waveguide dispersion.