**4. Results and discussion**

**•** Backscattering coefficients at C-band HH and VV pol, L-band HH and VV pol and the

Based on the field data, the integration ranges for Bayesian inference were selected with different values as is illustrated in the following part. The main aim of using different intervals was to test the sensitivity of the methods to prior information, Through these integrations, to each pixel a value of dielectric constant is associated, starting from the corresponding back‐ scattering coefficient values. Finally, with the formula proposed by [40] the dielectric constant values have been transformed to estimated values of soil moisture. The flowchart in Fig.8

combination of C and L band at HH pol for SMEX'02 test site.

52 Dynamic Programming and Bayesian Inference, Concepts and Applications

outlines the main steps of the algorithm, including training and test phase.

**Figure 8.** Flowchart of the Bayesian soil moisture approach applied to the Argentinean test site.

this specific algorithm is applied to SMEX'02 data.

As above mentioned, another version of the Bayesian algorithm was developed to take into account the effect of vegetation into the PDF. The flowchart of the algorithm is the same as shown in fig.8, but instead of Water Cloud Model there is an adaptation of the PDF mean to an empirical function related to VWC as detailed described in Notarnicola et al., 2007 [38]. The algorithm was developed to work with C, L and combination of C and L band. In this work, The main aim of the work is to verify the sensitivity of the algorithm to prior conditions of roughness and vegetation in order to optimize the accuracy of the results. Based on this concept several retrievals were performed for different conditions of surface roughness, with specific algorithms for each coverage type in the study area. In the following the results on the Argentinean and SMEX'02 test sites are presented and discussed.
