Two operational algorithms for the retrieval of snow depth and soil moisture content from AMSR-E data

In this paper two algorithms able to generate maps of snow water equivalent and soil moisture content have been developed and implemented within the framework of an Italian project devoted to the civil protection from floods and landslides. The two algorithms are based on Artificial Neural Network techniques and have been first tested on two Italian areas, one an agricultural area in the Po valley for the soil moisture algorithm and the second one on the Alps for the snow algorithm. After this test phase, the algorithm were validated on larger areas (the Po valley and Scandinavia), where some information on meteorological conditions and cartographic data were available. A method for the improvement of spatial resolution was also developed at IFAC, in order to obtain a resolution on ground at the lower frequencies comparable to the sampling rate (10km×10km). Due to the satisfactory results obtained on these areas, the two algorithms were subsequently validated at a worldwide scale, although, in this case, there was not availability of reference information or ground truth data. In any case, the global maps obtained resulted at least in agreement with the season and the meteorological and climatic conditions of the various zones.