Influence of Snow and Plant Covers on the Seasonal Radar Remote Sensing Signal Variations

Abstract

Listen

Translate article

In this paper are studied the radar images of land surface signal variations caused by seasonal changes in the properties of soil, snow and vegetation in the permafrost zone. For studies was used the Franklin Bluffs test site in Alaska, where are conducted the regular ground meteorological observations of the temperature and moisture conditions of soil. As the original data of remote sensing are used SAR low resolution images taken by the ERS-2 and JERS-1 instruments at C and L bands. A numerical simulation was performed, in a large-scale roughness approximation, for the radar backscatter relating to the rough soil covered by vegetation and snow. In the course of simulation, the refractive mixing dielectric model for the soil, snow, and vegetation was employed. The results of simulation were found to be in a good quantitative agreement to the backscatters derived with the use of radar images. The snow density was assumed to be uniform in depth, with the content of liquid water being changed in the course of melting. The complex dielectric constant (CDC) of snow in terms of its dependence on snow mass density and moisture was described by empirical model (3). For the soil CDCs, the generalized refractive mixing dielectric model (GRMDM) of (4) was employed, which takes into account the temperature dependences for the CDCs of free and bound soil water, with the range of freezing temperatures being included into consideration (5), (6). The properties of vegetation layer in the course of summer season are characterized by the thickness of layer and it's apparent CDC, which is calculated with the use of refraction mixing dielectric model taking into account the volumetric fractions of air and moist biomass contained in the vegetation layer. The simulation of radar backscatter was performed on the basis of Kirchhoff's approximation. Using the developed thus model, the radar backscatter was numerically analyzed through a full seasonal cycle taking into account variations in temperature and moisture content in the soil, snow, and vegetation. Roughness parameters for the soil, snow, and plant layers were assumed to be statistically independent. This simulation was conducted for L- and C- bands. The data on soil, snow, and air temperatures, along with the vegetation biomass, were acquired from the database given in (7). The values of temperatures used are presented in Table I. In the first part are presented soil surface temperatures and air, which correspond to the dates of the analyzed images in August- October of 1997. In the second part of the table assigned averaged on several days temperatures of soil surface of 1998, which are used for the analysis and the simulation in March- August.