Bistatic scattering and emissivities of random rough dielectric lossy surfaces with the physics-based two-grid method in conjunction with the sparse-matrix canonical grid method

Bistatic electromagnetic wave scattering from a two-dimensional (2-D) lossy dielectric random rough surfaces [three-dimensional (3-D) scattering problem] with large permittivity is studied. For media with large permittivities, the fields can vary rapidly on the surface. Thus, a dense discretization of the surface is required to implement the method of moment (MoM) for the surface integral equations. Such a dense discretization is also required to ensure that the emissivity can be calculated to the required accuracy of 0.01 for passive remote sensing applications. We have developed a physics-based two-grid method (PBTG) that can give the accurate results of the surface fields on the dense grid and also the emissivities. The PBTG consists of using two grids on the surface, the coarse grid and the required dense grid. The PBTG only requires moderate increase in central processing unit (CPU) and memory. In this paper, the numerical results are calculated by using the PBTG in conjunction with the sparse-matrix canonical grid (SMCG) method. The computational complexity and memory requirement for the present algorithm are O(N_scg log(N_scg)) and O(N_scg), respectively, where N_scg is the number of grid points on the coarse grid. Numerical simulations are illustrated for root mean square (rms) height of 0.3 wavelengths and correlation length of 1.0 wavelength. The relative permittivity used is as high as (17+2i). The numerical results are compared with that of the second-order small perturbation method (SPM). The comparisons show that a large difference in brightness temperature exists between the SPM and numerical simulation results for cases with moderate rms slope.