Surface identification using satellite microwave radiometers

The use of satellite microwave radiometers for identifying natural surfaces is analyzed. A retrieval technique is developed by considering the related mixed-pixel problem where two or more surfaces are contained within the viewing area. At a given frequency &ogr;, the emissivity measurement ε(&ogr;) depends on the fractional amounts f _n and a priori emissivities ε_n(&ogr;) where ε(&ogr;)=Σε_n(&ogr;)f_n. In applications involving surface identification the fractional amounts act as discriminants to identify the most likely surface among the a priori candidates. In principle, the fractional amounts can be obtained using multispectral measurements of emissivity. However, due to the limited spectral characteristics of emissivity the maximum number of distinguishable surfaces is reduced to three. The fractional amounts are derived using dual-frequency emissivity measurements and the effects of errors in measurement and a priori values are analyzed