Interpretation of GPR response through FDTD simulation and response surface method (RSM)

Some practical situations encountered in civil engineering are detection of buried pipes and manholes in soil or under pavements and characterization of soils at construction sites. Ground penetrating radar is an useful tool for this purpose. GPR data or a radargram is a resultant of interactions between GPR signals at a particular frequency and polarization and properties of host medium (ground) and objects buried there in, if any, physical dimensions, relative dielectric permittivity and conductivity. For the correct interpretation of GPR data, it is very crucial to understand the effect of every independent factor involved in its formation. These problems can be overcome by controlled large scale laboratory studies or computational simulation studies. Simulation studies are preferred as faster and more powerful computing resources are available. But still, the simulations involve solving of complicated Maxwell's equations repeatedly using FDTD. In the present work, a few typical construction site scenarios have been identified. Simulation studies have been carried out in order to know the influence of parameters like medium relative dielectric permittivity, object relative dielectric permittivity, pipe diameter, antenna frequency, conductivity etc and used to interpret real GPR data. Situations such as a pipe buried in soil (medium and object relative dielectric permittivities, pipe diameter and frequency are the variables) and void in soil medium (medium and void relative dielectric permittivities and void sizes are the variables) have been considered and GPR responses (amplitude) obtained. A statistical technique, namely, Response surface method (RSM), has been applied for establishing the relationship between input variables and the output (amplitude). The work shows that, simulation studies help in understanding the individual effect of all the factors which are responsible for GPR data and in turn help in interpreting real GPR data. For a given GPR scenario, using RSM, effect of influencing factors on amplitude can be understood.