Retrievals of low integrated water vapor using MIR and SSM/T-2 measurements

Satellite radiometric measurements at 150, 183.3+-3, and 183.3+-7 GHz have previously been used to retrieve integrated water vapor <1 g/cm/sup 2/ over Antarctica. The effects of the frequency dependence of surface emissivity and the variation of surface temperature on the retrieval, which have not been closely examined in the studies, are analyzed. Using four days of near-concurrent airborne and satellite radiometric measurements, it is shown that the previously derived retrieval algorithm could overestimate or underestimate integrated water vapor by up to 0.1 g/cm/sup 2/, depending on whether the surface emissivity increases or decreases with frequency. The average of the absolute value of the bias for each flight case studied is <0.04g/cm/sup 2/. Additionally, surface skin temperature is shown to vary substantially over a range from 240270 K during these four days of measurements; the corresponding effect on the retrieval of integrated water vapor is comparable to that due to frequency dependence on surface emissivity. The quantitative correction needed for this effect is dependent upon the magnitude of integrated water vapor. At high values of integrated water vapor of 0.6-0.8 g/cm/sup 2/, the corrections are as large as 0.1 g/cm/sup 2/ for changes of surface temperature of +-10 K. A simple procedure is implemented to correct for this error, which significantly improves the retrieval. Correction for the frequency dependence of surface emissivity is nontrivial when using currently available satellite measurements; in order to properly correct this effect, an additional channel of measurements, e.g., at 220 GHz, is required.