Development of a land surface emissivity algorithm for use by microwave rain retrieval algorithms

We have been developing a data-set of global land surface microwave emissivity calculated from 9-channel Brightness Temperatures (TBs) from the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) and atmospheric profile data from Japanese 25-year Reanalysis Project (JRA-25). The surface emissivity is derived using the non-scattering radiative transfer equation for regions identified as no-rain by TRMM Precipitation Radar (PR). An Empirical Orthogonal Function (EOF) analysis has been applied to this emissivity data-set. Emissivities at high frequencies, difficult to estimate due to high sensitivity to clouds and water vapor, are estimated from lower frequencies by using the principal components. Contributions from EOF1 to EOF4 are dominant and with the others being less than 1 %. Therefore, 5 high-frequency emissivities can be estimated from the other 4 emissivities at lower frequencies with 4 principal components. For example, when 37 GHz Horizontal emissivity on June 1998 is estimated from 4 channels of 10 and 19 GHz, correlation coefficient with the original estimate is 0.93 and the result of linear fitting shows an inclination of 0.97 and a cutoff of 0.02 for global data. This estimation method is applied for each area, each land surface condition (surface type and soil wetness) and so on, in search of optimal performance of the algorithm. The advantage of using the EOF analysis as described above is to minimize the cloud contamination at high frequency TB. A cloud-clearing method is also explored to improve the reliability of the EOFs.