Electromagnetic Modeling for Radar Remote Sensing of Snow-Covered Terrain

Abstract

This thesis investigates the radar remote sensing of snow-covered terrain for estimation of snow equivalent water on global scale. The importance and impact of this research stems from the fact that water from snowmelt is the major source of water for inland cities and agriculture during summer. This effort is focused on developing a physics-based model for snow and a fully coherent polarimetric scattering model for snow above ground. Both the physical model and the forward polarimetric scattering model present a significant improvement compared to the existing models for snowpack. Computer-generated snow media are constructed using 3-D spatial exponential correlation functions, along with Lineal-Path functions that serve to preserve the connectivity of the snow particles. A fully-coherent model is presented through the use of the Statistical S-matrix Wave Propagation in Spectral-Domain (SSWaP-SD) technique. The SSWaP-SD depends on the discretization of the medium into thin slabs. Several realizations of a thin snow slab are solved numerically to form the statistics of the scattering matrix representing such a thin snow layer. For each thin slab of the snow-pack, a corresponding polarimetric N-port (representing different directions of scattering) S-matrix is generated. These S-matrices are cascaded using the SSWaP-SD method to calculate the total forward and backward bistatic scattered fields in a fully coherent way. The SSWaP-SD, in conjunction with a Method of Moments (MoM) code based on the Discrete-Dipole Approximation (DDA), is chosen to leverage both the time-efficient computations of the DDA and the full-coherency of the SSWaP-SD method, simultaneously. In addition to the MoM-DDA, a Finite Element Method (FEM) based on commercial software is used for cross-comparison and validation. The simulation results of the backscattering from an arbitrary thick snow layer are presented and validated with measurements. The underlying rough ground surface response is then estimated through both an analytical technique based on the Physical Optics (PO) method and a numerical solver based on MoM using a commercial full-wave solver. Finally, the complete response is then calculated by cascading the S-matrices representing the snow and the rough surface responses. The simulation results of the backscattering are presented using a Monte-Carlo process, which show very good agreement with measurements.