Soil Moisture Estimation in Time With Airborne D-InSAR

Abstract

In this paper the potential to estimate soil moisture in time and in the presence of vegetation using differential SAR interferometry is explored. SAR interferometry (InSAR) is an established technique that allows the location of the effective scattering center within the resolution cell and the estimation of topographic height variations based on an interferogram formed by two images acquired at both ends of a spatial baseline. Concerning surface parameter estimation, there are two aspects that make SAR interferometry to be of interest. The first one is the potential sensitivity of interferometric observables to surface parameters. The second one is the potential of multi-parameter InSAR techniques to separate volume from surface scatterers. This is a challenge regarding the unsolved problem of surface parameter estimation in the presence of vegetation cover. Concerning now pure surface scatterers, different attempts have been reported in the literature to use the information content of the interferometric coherence and phase for the estimation of surface parameters such as surface roughness and volumetric soil moisture: The effect of roughness on the volume decorrelation term was addressed in [1] and an analytical expression under the Physical Optics (PO) approximation has been derived. In [2] and [3], the effect of temporal changes of soil moisture content on the interferometric observables has been discussed and evaluated using indoor experimental data. In [4], an analytical expression has been derived which relates surface roughness and volume decorrelation under the Kirchhoff approximation, and furthermore, the effect of in homogeneities in the dielectric properties of the surface in a sub-resolution scale has been addressed. These investigations demonstrated that the effect of surface roughness on spatial decorrelation is too weak to be of practical importance - for conventional InSAR configurations and realistic roughness scales. Indeed, as long as there is no significant penetration into the soil, natural surface scatterers are characterized by a localized (vertical) scattering center, and spatial decorrelation can be widely compensated by applying spectral range filtering. In this case, only temporal and SNR decorrelation affect the overall interferometric coherence. On the other hand, changes in soil moisture affect the vertical location of the phase center. But since the resulting height change is scaled by the baseline, the corresponding phase changes are also very small and become difficult to detect [2],[3]. A promising technique to overcome this problem is using differential interferometry which enables to measure small phase center changes related to soil moisture changes [5]. The main focus of this paper is to investigate this small phase changes and separate the interferometric phase change coming from the volume versus the one coming from the change of soil moisture [6]. The approach that we use is differential SAR interferometry at L-band and in a fully polarimetric mode [7],[8]. SAR polarimetry will allow us to decompose different scattering mechanisms and SAR interferometry will provide the phase center location. In addition surface scattering and a simple vegetation model are used to simulate and invert the soil moisture.