Radar retrieval of subsurface parameters for layered media with nonsmooth interfaces

The solution to the inverse problem for a three-layer medium representing a large class of natural subsurface structures is developed in this paper using radar data. The retrieval of the layered medium parameters is accomplished as a sequential nonlinear optimization starting from the top layer and progressively characterizing the layers below. The optimization process is accomplished by an efficient iterative technique built around the solution of the forward scattering problem. The forward scattering process is formulated by using the Extended Boundary Condition Method (EBCM) and constructing reflection and transmission matrices for each interface. These matrices are then combined into the generalized scattering matrix for the entire system, from which radar scattering coefficients are then computed. To be efficiently utilized in the inverse problem, the forward scattering model is simulated over a wide range of unknowns to obtain a complete set of subspace-based equivalent closed form models that relate radar cross section coefficients to the sought-for parameters including dielectric constants of each layer and separation of the layers. The inversion algorithm is implemented as a modified conjugate-gradient-based nonlinear optimization. It is assumed that multifrequency radar measurements are available from tower-mounted or airborne platforms, for example at typical radar frequencies of L-band and P-band (UHF). It is shown that this technique results in accurate retrieval of surface and subsurface parameters, even in the presence of noise.