Clear-Sky Longwave Irradiance at the Earth’s Surface—Evaluation of Climate Models

Abstract

An evaluation of the clear-sky longwave irradiance at the earth?s surface (LI) simulated in climate models and in satellite-based global datasets is presented. Algorithm-based estimates of LI, derived from global observations of column water vapor and surface (or screen air) temperature, serve as proxy ?observations.? All datasets capture the broad zonal variation and seasonal behavior in LI, mainly because the behavior in column water vapor and temperature is reproduced well. Over oceans, the dependence of annual and monthly mean irradiance upon sea surface temperature (SST) closely resembles the observed behavior of column water with SST. In particular, the observed hemispheric difference in the summer minus winter column water dependence on SST is found in all models, though with varying seasonal amplitudes. The analogous behavior in the summer minus winter LI is seen in all datasets. Over land, all models have a more highly scattered dependence of LI upon surface temperature compared with the situation over the oceans. This is related to a much weaker dependence of model column water on the screen-air temperature at both monthly and annual timescales, as observed. The ability of climate models to simulate realistic LI fields depends as much on the quality of model water vapor and temperature fields as on the quality of the longwave radiation codes. In a comparison of models with observations, root-mean-square gridpoint differences in mean monthly column water and temperature are 4?6 mm (5?8 mm) and 0.5?2 K (3?4 K), respectively, over large regions of ocean (land), consistent with the intermodel differences in LI of 5?13 W m?2 (15?28 W m?2).