Explaining Variations in Microwave Surface Emissivity from Passive Microwave Observations

The Global Precipitation Measurement (GPM) Microwave Imager (GMI) onboard the core satellite orbits at a 65 degree inclination, providing more observations over land areas whose underlying microwave surface emissivity is subject to wider variations. The emissivity at each channel is highly correlated, suggesting that characterization for one channel should not be done independently of another. Also, the surface emissivity can change very rapidly with the onset of precipitation. Principal component analysis was performed on the emissivity retrieved from a large set of global, over-land, multi-year, clear-scene A-Train observations (AMSR-E and MHS). It is shown that the 10-89 GHz emissivity spectrum can be reconstructed from the first three principal components, which in turn were estimated directly from the observed brightness temperatures. When these coefficients were applied to entire orbits of AMSR-E data, the reconstructed emissivity exceeded unity for regions associated with the presence of over-land precipitation. Examples are shown for cases that are further subsetted by the presence and amount of previous-time precipitation.