الفهرس 
Electromagnetic Scattering from Rough Surface: Properties and Methods of Analysis.
Incident Beam for Avoiding Edge ReflectionsOnly 14 pages are availabe for public view
 |  | from | 210 |  |  |
| | | from | 210 | | |
Abstract
Synthetic Aperture Radar (SAR) imaging of ground surface is of great importance in many civilian and military fields and, especially in the field of earth remote sensing. SAR images are subjected to subsequent data processing techniques for the purpose of the target detection and classification required in many applications. For the researcher working in the field of SAR image processing, the demand for a variety of images for various types of ground surface and under different climate conditions is an essential requirement. Real SAR images that satisfy such research requirements may not be available for most of the researchers. The solution is to obtain such images by SAR simulators, which are capable of generating realistic images for a specific area of the land if the geometrical and physical models are provided. Also, they allow understanding and interpreting the properties of scattering from rough surfaces which may be natural or man-made in a simple, inexpensive and fast way.
Fully polarimetric SAR system has four polarization channels to get the maximum information about the imaged land in the resulting images. Applications such as crop classification, soil moisture extraction and ship detection require fully polarimetric SAR images.
Many SAR imaging simulation approaches are based on evaluating the radar cross section (RCS) from the structure in the spectral domain with high spectral resolution. The obtained spectral domain solution is, then, transformed to spatial domain to form the SAR image in the range and azimuth directions. This approach may be complex and requires huge memory space. Moreover, the spectral domain approach is, practically, limited by the electrical size of the imaged scene and can be applied only at low frequencies.
The present thesis introduces fully polarimetric, high-resolution and wide-swath SAR imaging simulator taking into consideration the operational parameters of a real SAR system.
A random rough surface (RRS) model characterized by two statistical roughness parameters, root-mean-squared height h_rms and correlation length l_c is generated using the spectral domain method. A geometric model for the generated RRS is then constructed to be suitable for the electromagnetic scattering assessment. A volumetric cubic cell model is built for the application of FDTD method, whereas triangular mesh model is built for the application of GTD-RT approach.
At low and intermediate frequencies; the Finite Difference Time Domain (FDTD) method is used to assess the electromagnetic scattering from RRS models characterized by roughness parameters comparable to the wavelength. The angular distribution of the RCS of the wave scattered from RRS model has two components: specular and diffuse components. As the surface roughness increases the backscattering is increase, meaning that the surface becomes brighter in the simulated SAR image.
At high frequencies; the ray tracing (RT) technique is applied in conjunction with the geometrical theory of diffraction (GTD) to evaluate the scattering from the RRS model working at both microscopic and macroscopic scales depending on the relative dimensions of the roughness parameters of the RRS model. Multiple bounces of the incident rays up to the second bounce are considered during the application of the GTD-RT (second order GTD-RT). In the present study, the depolarization caused by the multiple bounce scattering rays is considered during scattering evaluations from rough surfaces. So, the co-polarized scattering elements (VV and HH) and the cross-polarized scattering elements (VH and HV) are obtained from PEC or lossy RRS models with different roughness parameters. The dependence of the polarization of the scattered field on the rough surface geometrical properties is studied by considering the surface curvature.
The results of the present work achieved by the FDTD and GTD show that the scattering from RRS is combination of coherent and non-coherent components. The angular distributions of the scattering elements are presented to show the dependency of the scattering on the roughness degree of the RRS model. As the surface roughness increase, the non-coherent component becomes more prominent and hence, the backscattering increase. The effect of the physical properties on the scattering from RRS is presented to show that as the dielectric constant and the conductivity of the surface decrease, the reflection coefficients decrease and the surface appears with less brightness in the simulated image.
The proposed SAR imaging system is used to generate realistic images for different rough surfaces with specific geometrical and physical models. An image smoothing process for removing speckles from the simulated SAR images is applied. Realistic images for deterministic surfaces are obtained by the proposed simulator with a high pixel resolution of 5 cm. By considering the surface curvature effect on scattering from RRS, the proposed simulator is sensitive to the geometrical properties of the generated surface. Also, wide swath fully polarimetric SAR image for a region of Suez Canal with the real geometrical and physical parameters of the selected area is obtained by the proposed simulator with a resolution of 500 m.