| Retrieval of vegetation and underlying ground surface parameters has become one of the major applications of microwave remote sensing. To accomplish this task, a necessary step is to construct high-fidelity scattering models by which the relationship between all target parameters to the radar backscatter can be established. Many scattering models have been proposed for both rough surfaces and vegetation canopies. However, many important issues such as the dielectric inhomogeneity of the underlying rough surfaces, and the near-field scattering interaction among the vegetation particles and rough surfaces, have not been addressed yet.; This dissertation provides electromagnetic scattering solutions for (1) slightly rough surfaces with inhomogeneous dielectric profiles and (2) rough surfaces covered with short branching vegetation. The rough surface scattering model presented in Chapter 2 is the first analytical model which includes surface scattering up to the second order and accounts for the dielectric inhomogeneity of dielectric medium. The scattering model for short branching vegetation is a polarimetric and fully coherent model which accounts for the near-field scattering interaction between vegetation particles and underlying rough surfaces (in Chapter 3) and among vegetation particles (in Chapter 4). The construction of the scattering model for short branching vegetation and the characterization of its performance are presented in Chapter 5. Soybean plants which possess the desired branching structure are chosen as the test targets. Realistic computer-generated vegetation structures, which are essential when considering the coherence effect of the vegetation structure and the scattering interaction among particles, are used in this model. The statistics of the polarimetric radar backscatter of vegetation medium is characterized by performing Monte-Carlo simulations.; The scattering models developed for individual components and the overall scattering model have been validated by conducting polarimetric radar backscatter measurements, using indoor measurement facilities, and the University of Michigan's truck-mounted polarimetric scatterometers. In all cases, good agreement is shown. Sensitivity analysis are performed as a guide for developing retrieval and target-classification algorithms. In addition to the aforementioned scattering models, a design procedure for a class of optimum corner reflectors suitable for the calibration of the imaging radars is also presented in Chapter 6. |
