Markov Chain Monte Carlo method applied to a Bayesian fusion of remotely sensed data for surface parameters retrieval

An algorithm is presented for retrieving soil parameters using microwave remotely sensed data. The algorithm is based on Bayes´ theorem of conditional probability and combines prior information on soil moisture and surface roughness with remote sensing measurements. In the Bayesian inference, the key point is the evaluation of a joint density probability function based on the knowledge of data sets consisting of soil parameters measurements and of the corresponding remote sensing data. The calculation of the marginal distribution has been obtained by a numerical integration known as Markov Chain Monte Carlo. This method is especially useful when the posterior density function has not a standard form. Furthermore, it is possible to obtain, at the same time, the distribution for all the parameters included in the process.