
Full text loading...
The finite-difference time-domain (FDTD) method has been adapted to accurately model GPR data. The method is based on explicit finite-difference approximations of Maxwell's curl equations. The model is set-up by dividing a finite-size volume into grid cells on the order of one-tenth of a wavelength in dimensions. Electric and magnetic field vectors are positioned along the'edges and normal to the sides of each grid cell. Specification ofthe electrical and magnetic properties for each grid cell permits modeling of coaxial feed cables, antennas, antenna enclosures, the air-gap between the antenna and the ground, and electrical and magnetic heterogeneity within the ground. During program execution, a voltage impulse is input in modeled balanced coaxial cables feeding the transmit antenna. The program is executed over the desired number of time-steps to obtain a full trace of data from modeled coaxial cables attached to a receive antenna. Special absorbing boundary conditions (ABCs) are used on the outer boundaries of the FDTD grid to keep energy impinging on the boundaries from reflecting back into the grid. Model results are compared to published field pattern data and measurements made over targets buried in the OSU GPR test pit. The absolute amplitude of FDTD modeled target reflection data is within 3.3 dB of data obtained from pit measurements. Both the frequency content and waveform characteristics ofthe modeled data also agree well with the experimental data.