Uncertainty Analysis of Surface Downward Longwave Radiation Models Based on Cloud Base Temperature

Cloud base temperature (CBT) is a crucial factor determining the surface downward longwave radiation (SDLR) under cloudy conditions. Theoretically, CBT-based parameterized models offer more accurate representations of cloud radiation effects and SDLR compared to simplistic models. However, they have poor performance in practical retrievals and have less development and application than other models. This study aims to pinpoint the shortcomings of existing CBT-based models, quantify model errors, and evaluate the impact of key parameter errors on SDLR retrieval results. Using simulated datasets based on radiative transfer models and ground-based remote sensing datasets, we conducted a detailed analysis of four CBT-based models. Our findings reveal that current model formulations inadequately capture the contributions of the atmosphere and cloud, leading to overestimation of the former and underestimation of the latter. However, these errors can partially offset each other. Under accurate parameter conditions, mean SDLR errors are within 10 W/m² for Diak, Gupta-Cal, and Wang models, and approximately −5 W/m² for the Schmetz model. The influence of cloud base height (CBH) and cloud fraction (CF) is significant and complex. When errors in CBH and CF are combined, CF error exerts a dominant influence. Surface downward longwave radiation error is insensitive to CBH error when CF is underestimated, while the impact of CBH error on SDLR estimation is notable when CF is overestimated. Regardless of CBH error, SDLR error is sensitive to CF error. Furthermore, when model errors are combined with cloud parameter errors, model errors may amplify or partially offset the impacts of parameter errors.