Abstract
Determination of the long wave (LW) atmospheric radiation (lambda=4-100 mu m) is of great importance for several applications. In this study, the effect of screen level and upper air temperatures and water vapour contents on the LW radiation for clear skies has been investigated using 4years of measurements conducted in Adelaide (34.9 degrees S;130.6 degrees E), South Australia. A two-variable model, which depends on the screen level temperature and vapour pressure, has been developed to calculate the LW radiation. The model can predict the hourly measured data with a correlation coefficient of 0.88, giving mean bias error (MBE) of 0.21W/m(2), root mean square error (RMSE) of 17.3W/m(2), and a mean percentage error (MPE) of -0.11% with respect to hourly observations. The performance of the multilinear model was tested against a 1-year independent data set. In this case, the MBE, RMSE, MPE, and correlation coefficient were 0.94W/m(2), -2.2W/m(2), 18.1W/m(2), -0.5%, and 0.85, respectively. Using radiosonde data, the total atmospheric water content has been calculated and its effect on the LW radiation has been investigated. A simple model that uses the screen level temperature and the precipitable water vapour (PWV) has been proposed. The correlation coefficient, MBE, and RMSE for this model were 0.93, 0.08W/m(2), and 12.12W/m(2), respectively; an MPE of 0.09% was reported by the model. The predictions of ten well-known clear sky models have been evaluated against the measured LW radiation data. These models were, then, adjusted and their performances have been assessed. The impact of the atmospheric mean weighted temperature on the LW radiation was studied. We have found that this temperature has less of an effect on the LW atmospheric radiation than screen level temperature. Finally, the clear sky model which uses the screen level parameters has been adjusted to account for the effect of clouds on the LW radiation. The model shows an acceptable predictability for the LW radiation under all sky conditions. The MBE, RMSE, MPE, and correlation coefficient for this model were 0.51W/m(2), 22.5W/m(2), 0.2% and 0.80, respectively.