Development of a Nonlinear Statistical Method for Estimating the Downward Longwave Radiation at the Surface from Satellite Observations

This paper develops a nonlinear statistical method that uses satellite radiance observations directly to estimate the downward longwave radiation (DLR) at the earth's surface, a necessary component of the surface energy budget. The proposed technique has rms regression errors of about 9 W m^sup -2 ^for clear-sky conditions, and about 4 to 8 W m^sup -2 ^ for overcast conditions, depending on the cloud levels. It is shown that this technique can produce unbiased estimates over a large range of meteorological conditions, which is crucial for climate studies. Sensitivity studies show that the DLR is most sensitive to errors in the cloud amount on average. Overall, the combined errors for an instantaneous DLR estimate, excluding the effects of the surface pressure errors, range from about 7 to 12 W m^sup -2 ^ when there is a +/-10% uncertainty in cloud amount and a+/-100 hPa uncertainty in cloud-base pressure. When the cloud amount uncertainty rises to 30%, the combined DLR error ranges from about 10 to 25 W m^sup 2 ^. This clear-sky DLR estimation technique was validated preliminarily by using simulated radiation data. The DLR differences between estimated and calculated values have a standard deviation of about 9 WM^sup -2 ^ and are unbiased in most conditions. The validity of the DLR estimation technique has been demonstrated; however, validation for cloudy conditions, comparison with surface observations, and improvements related to surface pressure dependence and skin temperature discontinuity are left for future study.