Retrieval of Ocean and Lake Surface Temperatures from Hyperspectral Radiance Observations

Abstract

This paper advances hyperspectral infrared (IR) radiative transfer techniques for retrieving water (ocean and lake) surface skin temperature from clear-sky radiance observations obtained within the longwave atmospheric window region (800?1000 cm?1). High spectral resolution has optimal potential for multispectral algorithms because of the capability to resolve, and thus avoid, gas absorption lines that otherwise obscure the surface signal in conventional narrowband radiometers. A hyperspectral radiative transfer model (RTM) is developed for varying satellite zenith angles, atmospheric profiles (cloud and aerosol free), surface wind speeds and skin temperatures, with atmospheric column transmittance spectra computed from fast models. Wind speed variations in surface emissivity and quasi-specular reflection are both rigorously accounted for. The RTM is then used for deriving retrieval algorithms based upon statistical and physical methodologies. The statistical method is based upon linear regression analyses of brightness temperatures, whereas the physical method is based upon solution of a linear perturbation form of the IR radiative transfer equation valid for window channels. The physical method is unique in its simplicity: It does not solve for atmospheric profiles, but rather relies upon local linearities about guess transmittances for extrapolating the skin temperature. Both algorithms are tested against independent forward calculations and then used to retrieve water surface skin temperatures from the National Polar-orbiting Operational Environmental Satellite System (NPOESS) Airborne Sounder Testbed-Interferometer (NAST-I) flown on board the NASA ER-2. The results demonstrate the capability of hyperspectral radiative transfer for providing an optimal correction for atmospheric gas absorption (viz., water vapor) from the new suite of environmental satellite IR spectrometers.