Evaluation of Methods to Estimate the Surface Downwelling Longwave Flux during Arctic Winter

Abstract

Surface longwave radiation fluxes dominate the energy budget of nighttime polar regions, yet little is known about the relative accuracy of existing satellite-based techniques to estimate this parameter. We compare eight methods to estimate the downwelling longwave radiation flux and to validate their performance with measurements from two field programs in the Arctic: the Coordinated Eastern Arctic Experiment (CEAREX) conducted in the Barents Sea during the autumn and winter of 1988, and the Lead Experiment performed in the Beaufort Sea in the spring of 1992. Five of the eight methods were developed for satellite-derived quantities, and three are simple parameterizations based on surface observations. All of the algorithms require information about cloud fraction, which is provided from the NASA–NOAA Television and Infrared Observation Satellite (TIROS) Operational Vertical Sounder (TOVS) polar pathfinder dataset (Path-P); some techniques ingest temperature and moisture profiles (also from Path-P); one-half of the methods assume that clouds are opaque and have a constant geometric thickness of 50 hPa, and three include no thickness information whatsoever. With a somewhat limited validation dataset, the following primary conclusions result: 1) all methods exhibit approximately the same correlations with measurements and rms differences, but the biases range from −34 W m−2 (16% of the mean) to nearly 0; 2) the error analysis described here indicates that the assumption of a 50-hPa cloud thickness is too thin by a factor of 2 on average in polar nighttime conditions; 3) cloud-overlap techniques, which effectively increase mean cloud thickness, significantly improve the results; 4) simple Arctic-specific parameterizations performed poorly, probably because they were developed with surface-observed cloud fractions whereas the tests discussed here used satellite-derived effective cloud fractions; and 5) the single algorithm that includes an estimate of cloud thickness exhibits the smallest differences from observations.