Comparison of Cloudy‐Sky Downward Longwave Radiation Algorithms Using Synthetic Data, Ground‐Based Data, and Satellite Data

Abstract

AbstractCloud plays a crucial role in surface downward longwave radiation (DLR). To determine optimal algorithms for cloud‐sky DLR calculation under various climates, three types of algorithms were assessed, (i) four empirical algorithms determining cloudy DLR by simple cloud correction using the cloud fraction, (ii) three parameterized algorithms determining the cloud contribution by cloud temperature, and (iii) a semiempirical algorithm, Zhou‐Cess, parameterized by the cloud water path. A sensitivity study was conducted using a Moderate Resolution Transmittance (MODTRAN) code to first determine the sensitive cloud parameters of DLR. Then, these algorithms were validated using synthetic data simulated by MODTRAN, ground‐observed data, and satellite‐observed data. When all the input parameters were accurate, the cloud‐correction algorithms showed poor performance. Cloud‐temperature‐based algorithms showed much better results but exhibited positive systematic biases. The Zhou‐Cess algorithm performed best but could not precisely describe the effect caused by cloud variations. When these algorithms were applied to ground‐ and satellite‐based data, the accuracies of DLR calculations were affected by the uncertainty in the atmospheric and cloud parameters. The simple empirical algorithms showed the poorest results. The cloud‐temperature‐based algorithms were greatly influenced by the uncertainty in cloud base temperature and cloud fraction and showed acceptable results when cloud fractions were accurate. The Zhou‐Cess algorithm revealed the best results at most sites and was less impacted by cloud parameter uncertainties; therefore, this algorithm is suggested for cloudy‐sky DLR calculation with poor‐quality cloud parameters.