Оцінка Параметрів Шорсткості Земної Поверхні За Двохполяризаційним Радіолокаційним Відбиттям Методом Зворотного Моделювання

Abstract

Radar remote sensing is a modern and advantageous method for aerospace research of the Earth. The mass commissioning of new high-resolution radar systems based on synthetic aperture radar (SAR) has greatly expanded the capabilities of radar imaging. The expediency of the SAR using was approved in domains that traditionally used remote sensing data, such as cartography, agriculture and forestry, mineral prospecting, environmental security, disaster monitoring, defense.The primary physical parameter of the land surface, which is registered by non-interferometric SAR, is the radar backscattering coefficient (sigma nought). The received radar signal is a source for the complex simulation of backscattering processes and evaluation of the secondary land surface physical and biophysical characteristics: texture, soil moisture, dielectric permittivity, vegetation cover structure, etc. Herewith the land surface roughness, described by the standard deviation of its vertical irregularities, is one of the most significant entities in any radar simulation. Therefore, the land surface roughness restoration using radar remote sensing data is a high-relevant task. The land surface roughness is an independent physical parameter, in much determining the radar backscattering. However, the roughness correlation length depends on the polarization of the radar signal.This paper describes a quantitative approach to the land surface roughness recovery by dual-polarization SAR imagery using separate measurements adjustment of independent physical value -dielectric permittivity in different polarizations. The proposed Baghdadi et al. semi-empirical calibration of IEM radar backscattering taking into account the polarization-dependent correlation length is used to ensure the physical equivalence of the land surface dielectric permittivity measurements in different polarizations. In addition, the paper provides the required computational equations, as well as the example of an actual Sentinel-1 radar image processing. The results obtained are generally correspond to the known physical patterns and landscape features of the study area.