Electronic Thesis and Dissertation Repository

Degree

Doctor of Philosophy

Program

Geography

Supervisor

Dr. Jinfei Wang

Abstract

In this thesis, novel methodology is developed to extract surface parameters under vegetation cover and to map crop types, from the polarimetric Synthetic Aperture Radar (PolSAR) images over agricultural areas. The extracted surface parameters provide crucial information for monitoring crop growth, nutrient release efficiency, water capacity, and crop production. To estimate surface parameters, it is essential to remove the volume scattering caused by the crop canopy, which makes developing an efficient volume scattering model very critical.

In this thesis, a simplified adaptive volume scattering model (SAVSM) is developed to describe the vegetation scattering as crop changes over time through considering the probability density function of the crop orientation. The SAVSM achieved the best performance in fields of wheat, soybean and corn at various growth stages being in convert with the crop phenological development compared with current models that are mostly suitable for forest canopy.

To remove the volume scattering component, in this thesis, an adaptive two-component model-based decomposition (ATCD) was developed, in which the surface scattering is a X-Bragg scattering, whereas the volume scattering is the SAVSM. The volumetric soil moisture derived from the ATCD is more consistent with the verifiable ground conditions compared with other model-based decomposition methods with its RMSE improved significantly decreasing from 19 [vol.%] to 7 [vol.%].

However, the estimation by the ATCD is biased when the measured soil moisture is greater than 30 [vol.%]. To overcome this issue, in this thesis, an integrated surface parameter inversion scheme (ISPIS) is proposed, in which a calibrated Integral Equation Model together with the SAVSM is employed. The derived soil moisture and surface roughness are more consistent with verifiable observations with the overall RMSE of 6.12 [vol.%] and 0.48, respectively.

Share

COinS