Inter-Sensor Comparison of Microwave Land Surface Emissivity Products to Improve Precipitation Retrievals

Microwave land surface emissivity acts as the background signal to estimate rain rate, cloud liquid water, and total precipitable water. Therefore, its accuracy can directly affect the uncertainty of such measurements. Over land, unlike over oceans, the microwave emissivity is relatively high and and varies significantly as surface conditions and land cover change. Lack of ground truth measurement of microwave emissivity especially on global scale has made the uncertainty analysis of this parameter very challenging. The present study investigates the consistency among the existing global land emissivity estimates from different microwave sensors. The products are determined from various sensors and frequencies ranging from 7 to 90 GHz. The selected emissivity products in this study are from the Advanced Microwave Scanning Radiometer for EOS (AMSR-E) by NOAA - Cooperative remote Sensing and Science and Technology Center (CREST), the Special Sensor Microwave Imager (SSM/I) by The Centre National de la Recherche Scientifique (CNRS) in France, TRMM Microwave Imager (TMI) by Nagoya University, Japan, and WindSat by NASA Jet Propulsion Laboratory (JPL). The emissivity estimates are based on different algorithms and ancillary data sets. This work investigates the difference among these emissivity products from 2003 to 2008 dynamically and spectrally. The similarities and discrepancies of the retrievals are studied at different land cover types. The mean relative difference (MRD) and other statistical parameters are calculated temporally for all five years of the study. Some inherent discrepancies between the selected products can be attributed to the difference in geometry in terms of incident angle, spectral response, and the foot print size which can affect the estimations. The results reveal that in lower frequencies (=<19 GHz) ancillary data especially skin temperature data set is the major source of difference in emissivity retrievals, while in higher frequencies (>19 GHz) the residuals of atmospheric effect on the signal cause inconsistency among the products. The time series and correlation between emissivity maps were analyzed over different land classes to assess the consistency of emissivity variations with geophysical variable such as soil moisture, precipitation, and vegetation.