A numerical investigation into the accuracy of determining dielectric properties and thicknesses of pavement layers using reflection amplitude GPR data

Absiruci-In processing GPR data from road surveys the
amplitudes of reflection data have often been used in obtaining
an estimate of the dielectric properties of pavement layers and
consequently an estimate of the layers' thicknesses. The values of
the dielectric constants of the layers are estimated using a
recursive procedure based on assumptions of plane wave
propagation, one dimensional target geometry and that all media
probed by the GPR are low-loss. In practice, this kind of data
processing requires the use of a calibration procedure in the field
which entails the use of a metal sheet placed on the surface of the
ground. This is used in order to provide a reference reflected
GPIl amplitude by assuming that the reflection coefficient of the
metal sheet is known. This reference information is then used in
the recursive determination of reflection coefficients of the layer
interfaces which are then used to obtain values for the dielectric
constants of the layers. Although, the assumption of one
dimlensional geometry (layered earth) appears to be a good
approximation for at least short sections of GPR road data the
other assumptions involved in establishing the processing
procedure may not be as valid and therefore they may have an
effect on the accuracy of the final layer thickness calculations.
This paper presents the results of a numerical investigation into
the accuracy of such processing procedure. Both, a two
dimensional and a three dimensional numerical simulators for
GPIt have been employed to investigate the accuracy with which
valuies of dielectric constants of layered models can be estimated
fronn simulated reflection amplitude GPR data. This relates to
the accuracy of the calculation of the layers' thicknesses which it
shoiild be noted that is the important engineering parameter that
needs to be determined by the GPR survey. Although, )he two
dimensional models employ a theoretical line source as a model
for the GPR's transducers the three dimensional ones employ a
complete model of a GPR antenna. The GPR simulators are
based on the finite-difference time-domain method.