Comparison of long‐wave radiation calculation methods over the United States

Several long‐wave radiation models have been developed which provide different methods of evaluating the emittance of the atmosphere. Most of these methods are derived for a location using the relationship provided by a researcher but utilizing local empirical coefficients. There are several, however, which are thought to be more universal in their predictive ability and these were tested on a data set collected in 1979, 1980, and 1981. Data were collected at 15 locations in the United States over a range in latitude from 26°13′N to 47°46′N and over an elevation range from −30 m to 3342 m, relative to sea level. The models evaluated in this study were the Brunt model with the original coefficients, the Swinbank, Idso‐Jackson, and Brutsaert models, each with two different empirical coefficients, and two recent methods proposed by Idso. Pooling all the information, the Brutsaert 1 model best described the data. However, with the Beartooth Pass and Estes Park data at elevations above 2500 m excluded, the Brunt, Idso‐2 and two Brutsaert relationships provided the best fit. In all cases the Swinbank and Idso‐Jackson equations, which correct for emittance solely on temperature, provided the poorest fit. An evaluation of the models for the 1981 season at Five Points, California, showed the Brunt, Brutsaert 1 and 2, and Swinbank models to best describe the data at that location. These analyses suggest that the original Brunt equation, the Brutsaert equation with a coefficient of 0.575, and the Idso 2 model would adequately predict long‐wave radiation amounts with errors less than 5% for most agricultural locations in the United States and that local correction for water vapor effects should not be necessary.