The inversion of the model was accomplished using at least two of the
three polarization-dependent models: HH–VV, HH–HV, VV–VH, and HH–
VV–HV. With such an approach, a priori knowledge of soil moisture is not
required in order to estimate LAI. The non-linear least square iterative algorithm
was stopped when the improvement between two consecutive
iterations was less than 10-6 or the number of iterations reached 400.
The model inversion was evaluated on the calibration dataset used to parameterize
the model, and subsequently on the validation dataset. The inversion
method is based on using initial values (i.e. first guess values) for
the output parameters and subsequently improving these initial values in
an iterative algorithm. In this case, initial values of 1.0 for LAI and 0.2 for
soil moisture were selected. The range of acceptable soil moisture estimates
was restricted to 0–0.55 m3 m-3, which is the valid range for