Adaptive bayesian algorithm for vegetated field parameters extraction by using multi-frequency and multi-polarimetric SAR images

This paper proposes an alternative inversion algorithm for extracting soil and vegetation parameters from multi-frequency, multi-polarimetric SAR data. The alternative approaches aim at identifying a useful modelization of the soil and vegetation response to radar signal and then indicate a possible solution to the extraction of soil moisture and vegetation water content. The core of the algorithm is based on the determination of probability density functions (pdfs) through a Bayesian methodology and has been initially developed for bare soils and tested on numerous data sets. The purpose is to apply this inversion algorithm to fields that have different levels of vegetation cover and considering different theoretical and empirical approaches. In fact, as already stated, a single approach, theoretical or empirical model, cannot be often applied on a wide number of cases. This paper addresses a possible solution based on two alternative different modelizations.