The biggest error when estimating the thickness of subsurface layering using GPR is not accounting for the dielectric differences of each material.
The speed of the radar wave will differ depending on the material in each layer especially when there is moisture. For example a wet granular may have a dielectric as high as 15 resulting in significant errors in the assumed layer thickness due to the slowing of the radar wave.
To alleviate this issue, some GPR manufacturers only use the surface reflection to determine a dielectric value and thus are only useful for calibration of the first asphalt surface. This does not work with subsequent asphalt or granular layers.
Most GPR manufacturers also make the assumption that the speed of the wave is constant when converting from the time recorded data to a depth profile and that the dielectric is 5 (2.23 times slower than the speed of light).
The PaveScout NM-GPR system uses a different approach. Each of the 8 transmitters sends a signal that is separately recorded by the multiple receivers in the system (normally done as 4 sets of 2 antennas). The result is multiple 3D arrays of GPR data, for each transmitter set. Using this data, in combination with the known distance between the transmitters it is possible to determine the dielectric value, or speed of the wave, for each of the layers in the GPR profile. There are two benefits to this method: