A comparison of the MIR-estimated and model-calculated fresh water surface emissivities at 89, 150, and 220 GHz

The airborne millimeter-wave imaging radiometer (MIR) measurements over three lakes (surface temperature /spl sim/273 K) in the Midwest region of the USA during February 1997 were used to estimate surface emissivities at 89, 150, and 220 GHz and the results were compared with those calculated from three different dielectric permittivity models for fresh water. The measurements were during clear and dry atmospheric conditions so that the column water vapor could be accurately retrieved and its effect on the MIR measurements predicted. The standard deviations of the estimated emissivities were found to be about 0.003, 0.004, and 0.008 for 89, 150, and 220 GHz, respectively. The errors of the estimation were calculated to be /spl plusmn/0.005, /spl plusmn/0.006, and /spl plusmn/0.011 in the same order of frequency, respectively, based on the MIR measurement accuracy of /spl plusmn/1 K in the brightness temperature range of 190-290 K. The estimated emissivities at normal incidence, under the assumption of a calm water surface, compare quite well with values generated by the model of P. Stogryn et al. (1995). These estimated values are slightly lower than those calculated from the model of H. J. Liebe et al. (1991) at both 89 and 150 GHz. The estimated 89 GHz emissivity is higher than that calculated from the model of W. Ellison et al. (1998). Additionally, the retrievals using different models of atmospheric absorption as well as off-nadir measurements of the MIR are explored. The impact of these retrievals on the comparison of estimated and calculated emissivities is discussed.