| Summary: | 碩士 === 國立中央大學 === 太空科學研究所 === 84 === The inversion of the random surface roughness and soil
moisture from radar backscattering coefficient is an
important subject in remote sensing. The relations
between the radar backscattering coefficient and surface
roughness and soil moisture are a highly nonlinear. In
this article, we first develop an inversion method based on
a soil scattering model and a dynamic learning neural
network, and then apply this method to the surface
parameters retrieval from SAR image. In order to
understand the relations between the soil dielectric
constant and moisture content, we perform a series of
measurements using two methods: dielectric probe and free
space, and examine the consistency of the two methods. The
interactions of electromagnetic wave and soil surface is
very complicated. Since the soil scattering model proposed
is accounted for single scattering from surface alone, this
is quite different from the real world condition. So we
apply the target decomposition theorem to single out the
single scattering term from the SAR image to improve the
inversion results.
|
|---|
