Efficient 3D forward modeling of GPR scattering from rough ground

Ground penetrating radar (GPR) is a viable tool for fast and high fidelity subsurface identification of buried landmines and improvised explosive devices (IEDs). The greatest barrier in GPR surveying systems is the obscuring effect of rough surface clutter. The finite difference frequency domain (FDFD) method can be used to calculate the scattering of the rough terrain. A full 3D FDFD analysis is required but is computationally burdensome when the GPR antennas are distant from the target region. We present an efficient hybrid forward-model to predict the scattering from randomly rough ground surface. In particular, this method models the additive clutter which can then be removed from the observed GPR signals. Our developed forward-looking model compares well with direct FDFD results. The developed model can be used readily for lossy and frequency-dispersive soils.