Estimation of the longwave radiation balance components: KUREX‐91 and FIFE studies∗

Abstract

Estimates of incoming longwave radiation flux densities (R1?) obtained using several empirical and theoretical algorithms were compared to flux densities measured during the FIFE‐87 and FIFE‐88 experiments. Statistical analysis showed that all models gave values that were greater than the measured values with biases ranging from 25 to 55Wm−2. The incoming longwave radiation models were corrected to reduce or eliminate these biases and applied to independent data sets obtained during FIFE‐89 and KUREX‐91. With these adjustments, the models showed significant improvement in their ability to accurately estimate Rl?. The algorithms of Brunt and Brutsaert performed the best of all models and gave consistently good results for the FIFE‐89 and KUREX‐91 studies. Compared to the Brunt equation, the Brutsaert equation had a higher d value and smaller random and systematic errors. Both models had mean relative errors of less than 2% for the KUREX‐91 study and about 5% for the FIFE‐89 study. Outgoing (emitted) longwave radiation (Rl?) flux densities were estimated from remotely sensed surface temperature data during FIFE‐89 and KUREX‐91. Compared to measured outgoing longwave radiation flux densities (Rl?), estimated Rl? values had mean relative errors of approximately 1% and 2% for KUREX‐91 and FIFE‐89, respectively.