Soil moisture inversion at L-band using a dual-polarization technique: a model-based sensitivity analysis
The inverse problem of estimating soil moisture from vegetated and non-vegetated terrain is investigated using a passive microwave technique. The radiometer response at L-band is modelled by the Peake's approach that involves the integration of bistatic scattering coefficients to calculate surface emissivity. For vegetated terrain, a discrete scatter model employing the distorted Born approximation is used to obtain bistatic scattering coefficients. For non-vegetated rough surface, the Kirchhoff 's rough surface approximation is used to calculate bistatic scattering coefficients. A series of vegetated and non-vegetated surfaces are simulated for this study by changing vegetation densities and ground roughness conditions. The forward model is simplified so that it can be inverted with a minimum of auxiliary information about the vegetation and the ground. An inversion procedure is set up such that the ratio of horizontal to vertical Fresnel reflectivity is expressed in terms of dual-polarized microwave brightness temperatures (available from the forward model or a radiometer). The reflectivity ratio expression is then solved to obtain soil moisture. In the case of weak vegetation such as soybean canopy, no information regarding vegetation is necessary in the inversion procedure because the difference between horizontal and vertical optical depth is assumed to be negligibly small at low to medium viewing angles. The consequence of this assumption on the soil moisture estimation from canopies of higher Leaf Area Index (LAI) is analysed. Similarly, in the case of non-vegetated rough terrain, inversion is carried out without information about rough surface because the rms surface roughness height cancels out in ratioing of the Fresnel reflectivity. To account for instrument and other errors in experimental data, a random noise is added to model data and the rms error in soil moisture inversion is calculated for vegetated and non-vegetated terrain. A comparison of the dual polarization with single (horizontal) polarization technique reveals that the dual polarization is a better technique for soil moisture estimation. The sensitivities and implications of a priori information are discussed in the context of developing a soil moisture estimation algorithm from satellites to be launched in the forthcoming space missions.