A preconditioned Quasi-Minimal Residual method for electrically large random surface scattering of remote sensing
A numerical model of wave scattering by means of the method of moments is an invaluable technique for the electromagnetic random surface scattering problems of remote sensing. However, this method is limited to electrically small bodies which are two-dimensional cases. The linear systems for electrically large bodies of three-dimensional cases are large and full, and shall be solved by more efficient algorithms. The iterative methods developed are suitable for large and full linear systems because they do not require the factorization of the matrix, A. The Helmholtz equations arise in many remote sensing problems and an iterative method, the Quasi-Minimal Residual method (QMR), is implemented to solve these systems. For the acceleration of the convergence to be effective, a preconditioned algorithm of QMR is developed. The efficiency and versatility of the numerical algorithm as a practical tool to study rough surface scattering of remote sensing is demonstrated.